Full text of "What Makes A Book Readble"

See other formats

133302

THE UNIVERSITY OF CHICAG(% ;
STUDIES IN LIBRARY SCIENCE*

WHAT MAKES A BOOK READABLE

THE UNIVERSITY OF CHICAGO PRESS
CHICAGO, ILLINOIS

THE BAKER & TAYLOR COMPANY
NEW YORK

THE CAMBRIDGE UNIVERSITY PRESS

LONDON

THE MARUZEN-KABUSHIKI-KAISHA

TOKYO, OSAKA, KYOTO, FUKUOKA, SENDAI

THE COMMERCIAL PRESS, LIMITED
SHANGHAI

WHAT MAKES A BOOK
READABLE

WITH SPECIAL REFERENCE TO ADULTS OF
LIMITED READING ABILITY

AN INITIAL STUDY

BY
WILLIAM S. GRAY, PH.D.

The University of Chicago
AND

BERNICE E. LEARY, PH.D.

St. Xavier College, Chicago

THE UNIVERSITY OF CHICAGO PRESS
CHICAGO ILLINOIS

COMPOSED AND PRINTED BV THE UNIVERSITY OF CHICAGO PRESS
CHICAGO, ILLINOIS, U.S.A.

PREFACE

IN A society like our own some ability to read is attained by
all but a small minority. This is the consequence of a long-
established belief that literacy is essential to intelligent
citizenry. Social enlightenment, personal advancement, enrich-
ment of experience, wholesome enjoyment of leisure all are
enhanced by the ability to read easily and understandingly
whatever interprets and illuminates the phenomena of life.

Unfortunately, however, a surprisingly large number of adults
are not interested in reading or are unable to find in the con-
tent of available material that which meets their particular in-
terests. Furthermore, about 50 per cent of our population cannot
read with ease and understanding much of the reading material
now available for adults. In recognition of the growing need to
serve more effectively this great mass of people, the American
Library Association and the American Association for Adult
Education have attacked the problem continuously since 1926
through such agencies as the joint Committee on the Reading
Interests and Habits of Adults and the Subcommittee on Read-
able Books. The activities of the former have included a sum-
mary of facts about adult reading, intensive studies of adult
reading interests and preferences, and searching inquiry into
the nature of adult reading materials, the results of which are
reported in part in this volume. The activities of the subcom-
mittee have been directed toward the preparation of lists of
readable books for adults of limited education. Among the con-
clusions reached by that committee is one of particular sig-
nificance at this time, namely, that many good books fall short
of being readable for the average reader in points that might
have been met quite easily. If such conclusions are valid, the
deficiency is doubtless due, in part at least, to lack of informa-
tion among authors and publishers relative to the factors which

vi PREFACE

make a book readable for different kinds of readers. It seems
apparent also that appropriate techniques are essential in de-
termining the right book for the right reader. With these con-
clusions and interpretations at hand, the study reported in this
volume was undertaken.

The purpose of the study is threefold: to make an initial
survey of current opinion concerning what makes a book read-
able for adults of limited reading ability; to study objectively a
small but important area of readability commonly designated
"ease" or "difficulty"; and to suggest possible applications of the
findings to the work of librarians in selecting the right book for
adult readers as well as to the task of writers and publishers in
preparing readable materials for different reading groups. That
we are keenly interested in the specific findings of this study and
their practical application is to be expected. We are equally
concerned, however, with the possibility of applying objective
procedures to the study of the wide range of additional problems
relating to readability. The present investigation has led to the
conviction that such procedures are both possible and practica-
ble, and what is even more important, probably, it has sug-
gested numerous possibilities of continued productive research
in this field.

In order to achieve the most valuable results in the further
study of readability, there is urgent need for active co-operation
of all who are professionally interested in the problem. Special-
ists in the field of adult education, librarians and readers' ad-
visers, authors and publishers, together with investigators who
have been studying special phases of readability, should share
experiences and findings, pool judgments, and define a broad
program of investigative activities which will attack systemati-
cally and thoroughly the various problems involved and make
practical application of the findings. Throughout the course of
such investigations, those to whom detailed responsibilities are
assigned should receive the constant counsel and guidance of
an advisory committee representing all the interests involved.
It is sincerely hoped that provision can be made in the near fu-

PREFACE vii

ture for such a co-ordinated and intensive study of the many
problems of readability that await solution.

Grateful acknowledgment is due to officers of the American
Association of Adult Education and of the American Library
Association for stimulating interest in this study; to the Carneg-
ie Corporation of New York for funds with which the study
was launched; to the Federal Emergency Relief Commission for
the assignment of research workers to the project associated
with the investigation; to the members of the Reading Habits
Committee of the American Association of Adult Education
and the American Library Association for their valuable help
and guidance throughout the study; to administrators and
teachers of adults and to parent-teacher organizations for sub-
stantial co-operation in the testing program; to Edgar Dale and
Ralph W. Tyler for invaluable technical assistance during the
major part of the investigation; to Michael West for critical
comments and stimulating suggestions; to Miss Frances Swine-
ford for generous statistical assistance; and to Louis R. Wilson,
Douglas Waples, Ralph W. Tyler, John Chancellor, Jennie M.
Flexner, and Doris Hoit for reading portions or all of the manu-
script and for constructive criticism.

WILLIAM S. GRAY
BERNICE E. LEARY

April 15, 1935

CONTENTS

PAGE

LIST OF TABLES . xi

LIST OF FIGURES xvii

CHAPTER

I. WHAT THE REPORT Is ABOUT i

II. WHAT Is A READABLE BOOK? 21

III. How WELL Do ADULTS READ? 57

IV. WHAT ELEMENTS INFLUENCE THE DIFFICULTY OF ADULT READ-
ING MATERIALS? 94

V. How Do ADULT READING MATERIALS DIFFER WITH RESPECT TO

ELEMENTS OF DIFFICULTY? 143

VI. WHAT Is THE DIFFICULTY OF ADULT READING MATERIALS? . . 194

VII. How TO SELECT READING MATERIALS FOR ADULTS .... 224

VIII. How TO PREPARE READABLE MATERIALS 261

APPENDIX

A. CHECK LIST OF POSSIBLE FACTORS OF READABILITY 293

B. FACTORS OF CONTENT, STYLE, FORMAT, AND ORGANIZATION RANKED

BY THREE GROUPS OF JUDGES 301

C. TECHNIQUE FOLLOWED IN CONSTRUCTING ADULT READING TESTS 311

D. AVERAGE READING SCORES ON ADULT READING TESTS .... 334

E. INTERCORRELATION OF FORTY-FOUR ELEMENTS 337

F. PREDICTED INDEXES OF DIFFICULTY OF THREE HUNDRED AND
FIFTY BOOKS 339

G. TEXTBOOKS USED IN INTERPRETING AREAS OF DIFFICULTY . . . 351
INDEX 353

LIST OF TABLES

TABLE PAGE

I. SUMMARY OF JUDGMENT CONCERNING THE RELATIVE IN-
FLUENCE ON READABILITY or THE FOUR MAJOR CATE-
GORIES 31

II. FACTORS RELATED TO CONTENT RANKED OF GREATEST AND
LEAST IMPORTANCE TO READABILITY BY EACH GROUP OF
JUDGES 33

III. SUMMARY OF JUDGMENT CONCERNING THE RELATIVE IN-
FLUENCE ON READABILITY OF GENERAL ASPECTS OF CON-
TENT 35

IV. FACTORS RELATED TO STYLE OF EXPRESSION AND PRESEN-
TATION RANKED OF GREATEST AND LEAST IMPORTANCE TO
READABILITY BY EACH GROUP OF JUDGES 37

V. SUMMARY OF JUDGMENT CONCERNING THE RELATIVE INFLU-
ENCE ON READABILITY OF GENERAL ASPECTS OF STYLE OF
EXPRESSION AND PRESENTATION 39

VI. FACTORS RELATED TO FORMAT RANKED OF GREATEST AND
LEAST IMPORTANCE TO READABILITY BY EACH GROUP OF
JUDGES 40

VII. SUMMARY OF JUDGMENT CONCERNING THE RELATIVE IN-
FLUENCE ON READABILITY OF GENERAL ASPECTS OF FOR-
MAT 42

VIII. FACTORS RELATED TO GENERAL FEATURES OF ORGANIZA-
TION RANKED OF GREATEST AND LEAST IMPORTANCE TO
READABILITY BY EACH GROUP OF JUDGES . 44

IX. SUMMARY OF JUDGMENT CONCERNING THE RELATIVE IN-
FLUENCE ON READABILITY OF GENERAL ASPECTS OF OR-
GANIZATION 45

X. FACTORS MENTIONED WITH GREATEST FREQUENCY BY
READERS AND READERS' ADVISERS IN COMMENTS REGARD-
ING THE READABILITY OF THE SAME BOOKS .... 50

XL NAME, GRADE-PLACEMENT, GEOGRAPHICAL LOCATION*, AND

SIZE OF GROUPS TESTED 63

XII. NEAREST AGE IN YEARS OF 756 ADULT SUBJECTS ... 65

XIII. LAST GRADE ATTENDED IN DAY SCHOOL, AS REPORTED BY

756 ADULT SUBJECTS 66

xii LIST OF TABLES

TABLE PAGE

XIV. OCCUPATIONS OF 756 ADULT SUBJECTS ...... 67

XV. NUMBER AND PERCENTAGE OF 756 ADULTS MAKING VARI-

OUS SCORES ON THE TESTS OF FICTION AND NON-FICTION . 68

XVI. MEAN SCORES FOR EACH OF TWELVE GROUPS ON THE

ADULT READING TESTS .......... 70

XVII. COEFFICIENTS OF CORRELATION BETWEEN FORM i AND

FORM 2 OF THE ADULT READING TEST ...... 73

XVIII. ADDITIONAL GROUPS TESTED ON THE MONROE STANDARD-

IZED SILENT READING TEST ......... 75

XIX. NUMBER AND PERCENTAGE OF ADULTS IN SIXTEEN REPRE-
SENTATIVE GROUPS MAKING VARIOUS GRADE SCORES ON
THE MONROE STANDARDIZED SILENT READING TEST . . 77

XX. READING ABILITY OF 756 ADULT SUBJECTS AS MEASURED BY

THE MONROE STANDARDIZED SILENT READING TEST . . 79

XXI. MEANS AND STANDARD DEVIATIONS OF THE SCORES OF
TWELVE ADULT GROUPS ON THE MONROE STANDARDIZED

SILENT READING TEST .......... 81

XXII. CORRELATIONS BETWEEN SCORES ON FORMS i AND 2 OF THE
ADULT READING TEST AND THE MONROE STANDARDIZED
SILENT READING TEST .......... 84

XXIII. NUMBER AND PERCENTAGE OF 151 VILLAGE AND RURAL
ADULTS MAKING VARIOUS GRADE SCORES ON THE MONROE
STANDARDIZED SILENT READING TEST ...... 87

XXIV. GRADE LAST ATTENDED AND GRADE SCORES IN ORAL READ-
ING AND IN COMPREHENSION FOR 75 VILLAGE AND 76 RURAL
ADULTS ............... 88

XXV. NUMBER AND PERCENTAGE OF ADULTS IN EACH OF THREE
NEGRO GROUPS MAKING VARIOUS SCORES ON THE MONROE
STANDARDIZED SILENT READING TEST ...... 90

XXVI. ELEMENTS OF EXPRESSION SUGGESTED AS POTENTIAL INDI-

CATORS OF DIFFICULTY IN ADULT READING MATERIAL . 98

XXVIL CORRELATIONS BETWEEN FORTY-FOUR POTENTIAL ELE-
MENTS OF DIFFICULTY AND THE AVERAGE SCORE OF EACH
TEST ITEM FOR THREE CLASSIFICATIONS OF ADULT SUB-
JECTS ................ 115

XXVIII. SIGNIFICANCE OF ELEMENTS OF DIFFICULTY FOR ALL READ-

ERS ................ iai

XXIX. SIGNIFICANCE OF ELEMENTS OF DIFFICULTY FOR "

READERS .............. 122

LIST OF TABLES xiii

TABLE PAGE

XXX. SIGNIFICANCE OF ELEMENTS OF DIFFICULTY FOR "POOREST"

READERS 123

XXXI. CORRELATIONS OF SIGNIFICANT ELEMENTS OF DIFFICULTY

WITH AVERAGE READING SCORE AND WITH EACH OTHER . 128

XXXII. INSTRUMENTS OF PREDICTION, COMBINING FOUR ELE-
MENTS OF DIFFICULTY 139

XXXIII. AVERAGE OCCURRENCE OF FIVE ELEMENTS IN Two SAM-
PLINGS OF TWENTY PASSAGES EACH FROM ONE BOOK . . 141

XXXIV. NAMES AND DATES OF ISSUE OF MAGAZINES USED IN THE
ANALYSIS AND THE ANNUAL CIRCULATION OF EACH . . 145

XXXV. NAMES OF FIFTEEN NEWSPAPERS SELECTED FOR ANALYSIS,

WITH CLASSIFICATION AND DATE OF ISSUE OF EACH . . 148

XXXVI. NAMES OF TWENTY-NINE BOOKS SELECTED FOR ANALYSIS,

WITH CLASSIFICATION AND AUTHOR OF EACH .... 149

XXXVII. SUMMARY OF DATA OBTAINED FROM AN ANALYSIS OF MONO-
SYLLABLES IN THE READING MATERIAL OF MAGAZINES,
NEWSPAPERS, AND BOOKS 152

XXXVIII. SUMMARY OF DATA OBTAINED FROM AN ANALYSIS OF PER-
CENTAGE OF DIFFERENT WORDS IN THE READING MATERIAL
OF MAGAZINES, NEWSPAPERS, AND BOOKS 154

XXXIX. SUMMARY OF DATA OBTAINED FROM AN ANALYSIS OF SYL-
LABIC SENTENCE-LENGTH IN THE READING MATERIAL OF
MAGAZINES, NEWSPAPERS, AND BOOKS 155

XL. SUMMARY OF DATA OBTAINED FROM AN ANALYSIS OF SIMPLE
SENTENCES IN THE READING MATERIAL OF MAGAZINES,
NEWSPAPERS, AND BOOKS 157

XLL PERCENTAGE OF MONOSYLLABLES IN THE READING MATE-
RIAL OF SIXTY-EIGHT GENERAL MAGAZINES .... 159

XLII. PERCENTAGE OF DIFFERENT WORDS IN SIXTY-EIGHT GEN-
ERAL MAGAZINES 167

XLIII. RANGE OF SYLLABIC SENTENCE-LENGTH AND MEDIAN SEN-
TENCE-LENGTH IN SIXTY-EIGHT GENERAL MAGAZINES . 172

XLIV. PERCENTAGE OF SIMPLE SENTENCES IN THE READING MA-
TERIAL OF SIXTY-EIGHT GENERAL MAGAZINES .... 174

XLV. RELATIVE DIFFICULTY OF GENERAL MAGAZINES BY COM-
POSITE CLASSIFICATION ON THE BASIS OF FOUR ELEMENTS
OF DIFFICULTY 176

XLVI. PERCENTAGE OF MONOSYLLABLES IN THE READING MA-
TERIAL OF FIFTEEN NEWSPAPERS 178

xiv LIST OF TABLES

TABLE PAGE

XL VII. PERCENTAGE OF DIFFERENT WORDS IN FIFTEEN NEWS-
PAPERS 179

XL VIII. RANGE OF SYLLABIC SENTENCE-LENGTH AND MEDIAN SEN-
TENCE-LENGTH IN FIFTEEN NEWSPAPERS 1 80

XLIX. PERCENTAGE OF SIMPLE SENTENCES IN THE READING MA-
TERIAL OF FIFTEEN NEWSPAPERS 181

L, PERCENTAGE OF MONOSYLLABLES IN THE READING MA-
TERIAL OF TWENTY-NINE BOOKS 184

LL VOCABULARY DIVERSITY IN TWENTY-NINE BOOKS . . . 187

LII. RANGE OF SYLLABIC SENTENCE-LENGTH AND MEDIAN SEN-
TENCE-LENGTH IN TWENTY-NINE BOOKS 189

LIIL PER CENT OF SIMPLE SENTENCES IN THE READING MATERI-
AL OF TWENTY-NINE BOOKS 190

LIV. BOOKS DEFINED AS RELATIVELY EASY, AVERAGE, OR DIF-
FICULT BY THE OCCURRENCE OF EIGHT ELEMENTS OF DIF-
FICULTY 192

LV. FORM USED IN RECORDING THE OCCURRENCE OF ELEMENTS
OF DIFFICULTY AND THE CALCULATION OF THE AVERAGE
READING SCORE 198

LVL DISTRIBUTION OF AVERAGE READING SCORES PREDICTED

FOR THREE HUNDRED AND FIFTY BOOKS 201

LVIL DISTRIBUTION OF AVERAGE READING SCORES PREDICTED

FOR EIGHTY-ONE TEXTBOOKS IN READING 216

LVIII. AVERAGE GRADE ACHIEVEMENT IN READING BY GROUPS
CLASSIFIED ACCORDING TO THE GRADE REACHED IN PUBLIC
SCHOOL 241

LIX. AVERAGE OCCURRENCE OF FIVE ELEMENTS OF DIFFICULTY

IN HUNDRED-WORD SAMPLINGS OF TWO VERSIONS OF

Treasure Island 266

LX. AVERAGE OCCURRENCE OF SEVEN ELEMENTS OF DIFFI-
CULTY IN HUNDRED-WORD SAMPLINGS OF TWO VERSIONS OF

Moby Dick 267

LXL AVERAGE OCCURRENCE OF SEVEN ELEMENTS OF DIFFICULTY

IN HUNDRED-WORD SAMPLINGS OFTwO VERSIONS OF Rotin-

son Crusoe 268

LXIL AVERAGE OCCURRENCE OF SEVEN ELEMENTS OF DIFFICULTY

IN HUNDRED-WORD SAMPLINGS OF TWO VERSIONS OF SilaS

Marner 268

LIST OF TABLES xv

TABLE PAGE

LXIII. AVERAGE OCCURRENCE OF FIVE ELEMENTS OF DIFFICULTY

IN HUNDRED-WORD SAMPLINGS OF TWO VERSIONS OF LeS

Miserable* 269

LXIV, AVERAGE OCCURRENCE OF FIVE ELEMENTS OF DIFFICULTY

IN HUNDRED-WORD SAMPLINGS OF TWO VERSIONS OF The

Vicar of Wakefield 269

LXV. TENTATIVE STANDARDS FOR USE IN WRITING FOR ADULTS

OF LIMITED READING ABILITY 288

LXVI. FACTORS RELATED TO CONTENT RANKED OF GREATEST AND

LEAST IMPORTANCE TO READABILITY BY LIBRARIANS . . 301

LXVII. FACTORS RELATED TO CONTENT RANKED OF GREATEST AND

LEAST IMPORTANCE TO READABILITY BY PUBLISHERS . . 302

LXVIII. FACTORS RELATED TO CONTENT RANKED OF GREATEST AND
LEAST IMPORTANCE TO READABILITY BY OTHERS INTEREST-
ED IN ADULT EDUCATION 302

LXIX. FACTORS RELATED TO STYLE OF EXPRESSION AND PRESEN-
TATION RANKED OF GREATEST AND LEAST IMPORTANCE TO
READABILITY BY LIBRARIANS 303

LXX. FACTORS RELATED TO STYLE OF EXPRESSION AND PRESEN-
TATION RANKED OF GREATEST AND LEAST IMPORTANCE TO
READABILITY BY PUBLISHERS 304

LXXI. FACTORS RELATED TO STYLE OF EXPRESSION AND PRESEN-
TATION RANKED OF GREATEST AND LEAST IMPORTANCE TO
READABILITY BY OTHERS INTERESTED IN ADULT EDUCA-
TION 305

LXXII. FACTORS RELATED TO FORMAT RANKED OF GREATEST AND

LEAST IMPORTANCE TO READABILITY BY LIBRARIANS . . 306

LXXIII. FACTORS RELATED TO FORMAT RANKED OF GREATEST AND

LEAST IMPORTANCE TO READABILITY BY PUBLISHERS . . 307

LXXIV. FACTORS RELATED TO FORMAT RANKED OF GREATEST AND
LEAST IMPORTANCE TO READABILITY BY OTHERS INTER-
ESTED IN ADULT EDUCATION 308

LXXV. FACTORS RELATED TO GENERAL FEATURES OF ORGANIZA-
TION RANKED OF GREATEST AND LEAST IMPORTANCE TO
READABILITY BY LIBRARIANS 309

LXXVI. FACTORS RELATED TO GENERAL FEATURES OF ORGANIZA-
TION RANKED OF GREATEST AND LEAST IMPORTANCE TO
READABILITY BY PUBLISHERS 309

xvi LIST OF TABLES

TABLE PAGE

LXXVII. FACTORS RELATED TO GENERAL FEATURES OF ORGANIZA-
TION RANKED OF GREATEST AND LEAST IMPORTANCE TO
READABILITY BY OTHERS INTERESTED IN ADULT EDUCA-
TION 310

LXXVIII. RELIABILITY OF COMPOSITE JUDGMENT FOR ITEMS REPRE-
SENTING THE RANGE OF VARIABILITY IN FORM i AND
FORM 2 322

LXXIX. CORRELATIONS BETWEEN SCORES ON FORMS i AND 2 OF THE
ADULT READING TEST AND THE MONROE STANDARDIZED
SILENT READING TEST 328

LXXX. VALIDITY OF THE ITEMS IN FORMS i AND 2 OF THE ADULT

READING TEST 330

LXXXI. RELIABILITY COEFFICIENTS, STANDARD DEVIATIONS, PROB-
ABLE ERRORS OF SCORES, AND RATIOS BETWEEN THE PROB-
ABLE ERRORS AND STANDARD DEVIATIONS OF SCORES ON
ADULT READING TESTS 332

LXXXII. CRITERIA OF DIFFICULTY FOR ITEMS ON FORM i, REPRE-
SENTED BY THE AVERAGE SCORE OF ALL READERS, "BEST"
READERS, AND "POOREST" READERS 335

LXXXIII. CRITERIA OF DIFFICULTY FOR ITEMS ON FORM 2, REPRE-
SENTED BY THE AVERAGE SCORE OF ALL READERS, "BEST"
READERS, AND "POOREST" READERS 336

LXXXIV. CORRELATIONS OF ELEMENTS OF EXPRESSION WITH AVER-
AGE READING SCORES AND WITH EACH OTHER . . facing 338

LXXXV. PREDICTED DIFFICULTY OF THREE HUNDRED AND FIFTY

BOOKS 339

LXXXVI. TEXTBOOKS IN READING USED IN THE INTERPRETATION OF
AREAS OF DIFFICULTY REPRESENTED BY THREE HUNDRED
AND FIFTY ADULT BOOKS 351

LIST OF FIGURES

FIGURE PAGE

1. OPINION CONCERNING THE INFLUENCE OF CLASSIFIED FACTORS ON
READABILITY 31

2. AN EVALUATION OF THE GENERAL ASPECTS OF CONTENT ... 36

3. AN EVALUATION OF GENERAL ASPECTS OF STYLE 39

4. AN EVALUATION OF GENERAL ASPECTS OF FORMAT 43

5. AN EVALUATION OF GENERAL ASPECTS OF ORGANIZATION ... 46

6. MEDIAN AND FIRST AND THIRD QUARTILES ON FORMS i AND 2 OF
THE ADULT READING TEST FOR TWELVE GROUPS OF ADULTS . . 71

7. SCORES OF 756 ADULTS ON THE MONROE STANDARDIZED SILENT
READING TEST 80

8. PERCENTAGE OF ADULTS BELOW SEVENTH-GRADE LEVEL OF READ-
ING ACHIEVEMENT AS DETERMINED BY THE MONROE TEST ... 82

9. SCHOOL GRADE AND READING GRADE ACHIEVEMENT .... 85

10. DISTRIBUTION OF "B" SCORES OF 75 VILLAGE AND 76 RURAL ADULTS
ON THE STANDARDIZED ORAL READING PARAGRAPHS AND THE MON-
ROE STANDARDIZED SILENT READING TEST 89

n. PERCENTAGE OF MONOSYLLABLES IN 1,000 WORDS IN THE READING

MATERIAL OF FIVE SELECTED MAGAZINES 161

12. PROPORTIONATE OCCURRENCE OF MONOSYLLABLES AND POLYSYL-
LABLES IN MAGAZINES OF GENERAL FICTION 163

13. PROPORTIONATE OCCURRENCE OF MONOSYLLABLES AND POLYSYL-
LABLES IN MAGAZINES OF GENERAL INFORMATION 164

14. PROPORTIONATE OCCURRENCE OF MONOSYLLABLES AND POLYSYL-
LABLES IN MAGAZINES OF WOMAN'S INTERESTS 165

15. ANALYSIS OF THE FIRST 175 SENTENCES IN THE SCO-SENTENCE
UNIT SELECTED FROM THE READING MATERIAL ov Aces . . . 169

16. ANALYSIS OF THE FIRST 175 SENTENCES IN THE SCO-SENTENCE UNIT
SELECTED FROM THE READING MATERIAL OF Review of Reviews . . 170

17. SYLLABIC LENGTH OF SENTENCES IN MAGAZINES OF GENERAL FIC-
TION 173

1 8. SYLLABIC LENGTH OF SENTENCES IN MAGAZINES OF GENERAL IN-
FORMATION 173

xviii LIST OF FIGURES

FIGURE PAGE

19. PROPORTIONATE OCCURRENCE OF MONOSYLLABLES AND POLYSYL-
LABLES IN DAILY NEWSPAPERS 182

20. SYLLABIC LENGTH or SENTENCES IN DAILY NEWSPAPERS . . . 182

21. PROPORTIONATE OCCURRENCE or MONOSYLLABLES AND POLYSYL-
LABLES IN POPULAR NOVELS 185

22. PROPORTIONATE OCCURRENCE OF MONOSYLLABLES AND POLYSYL-
LABLES IN POPULAR GENERAL BOOKS 186

23. AVERAGE READING SCORES OR INDEXES OF DIFFICULTY PREDICTED

FOR 350 GENERAL ADULT BOOKS 203

24. DEFINITION OF FIVE AREAS OF DIFFICULTY REPRESENTED BY THE
ADULT BOOKS STUDIED 207

25. PREDICTED DIFFICULTY OF READING TEXTS FOR ELEMENTARY AND
JUNIOR HIGH SCHOOLS IN TERMS OF RANGE, MEAN, AND STAND-
ARD DEVIATION 217

26. INTERPRETATION OF AREAS OF DIFFICULTY IN TERMS OF DIFFICUL-
TY REPRESENTED BY TEXTBOOKS FOR GRADES II TO IX . . . 218

27. OCCURRENCE OF DIFFERENT HARD WORDS IN READING MATERIALS

AT SUCCESSIVE AREAS OF DIFFICULTY 289

28. OCCURRENCE OF PERSONAL PRONOUNS IN READING MATERIALS AT
SUCCESSIVE AREAS OF DIFFICULTY ; 289

29. AVERAGE SENTENCE-LENGTH IN READING MATERIALS AT SUCCES-
SIVE AREAS OF DIFFICULTY 290

30. OCCURRENCE OF DIFFERENT WORDS IN READING MATERIALS AT
SUCCESSIVE AREAS OF DIFFICULTY 290

31. OCCURRENCE OF PREPOSITIONAL PHRASES IN READING MATERIALS

AT SUCCESSIVE AREAS OF DIFFICULTY 291

32. TABULATING CARD FOR RECORDING INDIVIDUAL TEST SCORES . 334

CHAPTER I
WHAT THE REPORT IS ABOUT

HOW much a person reads and what he reads undoubted-
ly are determined by many factors. Some pertain to
the reader to his proficiency in reading, to his mo-
tives for reading^ and to his reading interests and tastes. Others
relate to the reading materials to their accessibility and to
their readability. The co-ordination of these two sets of factors
for the purpose of getting the right book into the hands of the
right reader should go far toward extending and improving read-
ing habits.

But what is the right book? In other words, what makes a
book readable for a particular reader? This is the general ques-
tion with which we are concerned. Yet in this report we have
not answered the question completely. Nor have we attempted
to do so. What we have done is, first, to discover the trend of
opinion among librarians, publishers, and teachers and direc-
tors of adult education concerning the factors which they be-
lieve make a book readable. We have used the findings of this
preliminary inquiry to point the way to the major part of the
investigation, which is concerned with two problems. One is to
ascertain what elements in reading material make it easy or
difficult for adults, when the purpose of reading is defined some-
what narrowly; and the other, to discover how these elements
may be used in a more accurate estimate of the difficulty of
reading material of a similar sort. Throughout the report we
have attempted to indicate the types of subsequent research
needed to determine what is a readable book, and, therefore,
what is the right book for a particular reader.

THE PROBLEM IS A TIMELY ONE

There is a conspicuous interest at present in defining, pre-
paring, and selecting readable materials. This may be explained

2 WHAT MAKES A BOOK READABLE

in part by the fact that never before have readable books had
such a wide potential audience. Many forces are at work to in-
crease this audience. Shortening of the working day and the
working week, technological unemployment, and a prolonged
economic crisis have combined to create an increasing leisure
which may be considered potential reading time. Adult educa-
tion as an organized movement is giving a new sense of direc-
tion to the intellectual life of adults. Clubs, forums, councils,
and discussion groups are increasing rapidly. Five times as
many organized adult groups exist now as in 1929. Social and
recreational programs are being supplemented by discussions on
controversial issues of social, economic, and political import.
Frequently these programs go no farther than the presentation
and discussion of a timely topic. Hence they may fail to lead
the group to the larger understandings and deeper apprecia-
tions to be gained through the educational experience of sys-
tematic reading and study. Other programs aim to bring about
a more intelligent grappling with vital problems through an un-
derstanding of the conditions and forces creating them. To this
end related reading courses are outlined to fit the needs and in-
terests of the group. Co-operation is secured from librarians in
giving publicity to books bearing on the problem discussed at
group meetings. Reading for understanding steadily receives
definite encouragement and systematic guidance.

With the New Deal committed to the policy of using its
powers to alleviate economic distress, adult education is being
utilized as an important step toward recovery. Emergency pro-
grams are springing up on every hand. The unemployed are
being drawn into the classroom in the role of teacher or student.
A variety of educational activities are being rapidly initiated.
These include teaching native- and foreign-born adult illiterates
to read and write English; training persons who are physically
handicapped to do some remunerative work; giving vocational
training of a new type to adults whose specialized trade is no
longer in demand; and extending the general education of per-
sons who are dissatisfied with their past attainments. In the

WHAT THE REPORT IS ABOUT 3

majority of cases adult students are being guided to find a way
out of an economic situation they but partially understand. In-
asmuch as the "way out" may lie in the command and interpre-
tation of facts which explain progressive changes and evolving
social life in a dynamic world, the place and purpose of read-
ing cannot be disregarded.

But more important than the increase in the number of po-
tential readers is the evidence that a new vigor is actually being
developed in systematic reading. This is inspired by a desire
for economic security, cultural advancement, or social adjust-
ment. Current demands for reading are singularly common.
Readers at all levels of learning are asking for more information
about their own vocations and professions information that
will add to their efficiency and tend to create in them a feeling
of security.

Few of us can now be content to be specialists in a single
field. Our complacency has met uncomfortable jolts as rapid
changes in industrial processes and precipitous shifts in social
aims and outlooks have outdistanced our understanding or ap-
preciation. As a consequence, we are asking for bodies of infor-
mation bearing on unfamiliar fields general information that
will give a comprehensive survey of a subject or a penetrating
insight into some of its aspects without a bewilderment of tech-
nical detail. Escape from reality or interpretation of reality in
a good book of fiction, travel, biography, or history is a part of
the pleasure every reader hopes to find in reading. Hence, books
for this purpose are also a part of the common demand. Ma-
terials presenting reliable information on current economic, so-
cial, and civic problems are being sought not only by the econo-
mist and sociologist but by the ordinarily indifferent layman
who has come to want something more than passing propa-
ganda on which to base his hopes and his decisions.

Evidence for this new vigor in reading lies in recent reports
from public libraries to the effect that library patronage is in-
creasing rapidly in quantity and quality. In quantity, because
enforced leisure, prolonged idleness, and participation in edu-

4 WHAT MAKES A BOOK READABLE

cational activities all serve as an impetus for immediate reading
and study; in quality, because the lifetime reader, who has
already acquired the habit of reading for information or for
pleasure, is now forced by economic circumstances not only to
read more but to borrow rather than to buy the books he reads.

Writers and publishers are making vigorous effort to meet
increasing reading demands by preparing readable materials
adapted to the varied needs and interests of the general reader.
Probably this effort is more consciously directed toward books
of general information than toward fiction. This may be ex-
plained -in part by the interrelation of two circumstances. One
is the growing interest manifested by adults in non-fiction, es-
pecially non-fiction that carries a flavor of fiction or promises
some help in solving a personal problem. Lists of best-sellers
tend increasingly to include these types of non-fiction as well as
the more serious type of fiction. The second circumstance is the
patent need for more informational, non-technical material
within the understanding of the reader.

Convincing examples of publishers' efforts to meet new needs
and interests are to be found in semi-narrations of history,
geography, and other of the sciences, and in brochures of basic
information pertaining to social, economic, and political prob-
lems. For whom are these materials readable? Their popularity
is evidence that among certain classes of readers they are satis-
fying a need for non-technical information about technical
themes. It seems reasonable to assume that they may be read-
able for the person of average or more than average reading ex-
perience and yet be altogether unreadable for the adult who has
read few, if any, books, and who has acquired only a fair ability
to read. For there is no denying the fact that the reading public
is conspicuously stratified when its members are classified with
respect to how well they read. What materials are of appropri-
ate difficulty for readers at each level is one of the things this
report aims to show.

The evidence to be presented later supports the testimony of
librarians and teachers of adult classes that much reading ma-

WHAT THE REPORT IS ABOUT 5

terial of general adult interest is suitable only for readers at the
top stratum. For readers at considerably lower strata of read-
ing experience such material is difficult and abstruse. If, there-
fore, these persons do a meager amount of reading of informa-
tional non-fiction, one explanation may lie in a lack of mate-
rial adapted to their needs. If they turn to Dream World or
Ranch Romances for their fiction, it may be because relatively
little else of equal simplicity is available. For the near-illiterate
of still lesser reading ability, reading matter is even more limit-
ed. The seriousness of this condition has been emphasized by
objective evidence showing that the chief handicap to increasing
the reading efficiency of new literates lies more often in a lack
of readable materials than in serious disability of the learners.

But how can we know whether a book is readable for a par-
ticular reader? When we have the answer to this question we
shall have the secret of meeting current reading demands, of
getting the right book into the hands of the right reader,, and
ultimately of extending and improving reading habits.

THE PRESENT APPROACH

As stated at the outset, the present investigation is but an
initial approach to the definition of a readable book. Therefore,
as we outline briefly what this report contains, we shall indi-
cate also what it does not contain.

When we ask whether or not a book is readable, we meet a
counter question, Readable for whom ? From these two emerge
most, if not all, of the issues which at the present time seem to
be involved in the total problem of readability. The first directs
our attention toward the reading material and the qualities
which presumably affect its readability. It is with this question
that the present report is primarily concerned. With the second,
we face about and look at the reader for whom a book must be
satisfying if it is readable. In this case, we have in mind a par-
ticular class of reader one of limited reading ability.

Who is a reader of limited ability? He cannot be identified in
terms of years, for his age may range anywhere from sixteen to

6 WHAT MAKES A BOOK READABLE

ninety, probably. Neither can he be recognized by his occupa-
tional interests, for, unfortunately, the limited reader is repre-
sented in practically every type of occupation. His educational
experience, too, covers a wide range, from little or none to a
considerable amount. At least one feature probably character-
izes a limited reader. This is his inability to read with pleasure
and understanding any but the simplest adult materials, usually
cheap fiction or graphically presented news of the day. But how
are we to know that the reader of such materials is able to read
no better or that the person who reads not at all is unable to
read? These are important questions for which we need answers
if we hope to improve reading habits.

In this report we have considered the reader primarily from
the point of view of his ability to read certain kinds of materials
for certain specified purposes. In studying how well he reads,
in preference to what he wants to read or what he does read or
how much he reads, we do not mean to imply that ability is
more important than taste or interest or any other quality in
determining whether a particular reader will find a particular
book readable. We do maintain, however, that a reader's ability
is of great importance in determining how much difficulty he
will meet in reading materials prepared for him. And diffi-
culty is the aspect of readability we are studying in greatest
detail.

Yet even our study of reading ability is not complete. It is
commonly conceded that there is no such quality as general
reading ability. There is, rather, a series of specific abilities
which a person manifests in reading different kinds of materials
for different purposes. Presumably the ability in each case is
influenced by the degree of interest he has in the content, by the
nature of the outcome desired, as well as by a variety of other
factors. A true measure of reading ability, therefore, should be
considerably more comprehensive than the measure we have
obtained through the use of materials of our own selection, read
for the purposes which we have chosen to define. Additional
study should be made to discover the degree of relationship be-

WHAT THE REPORT IS ABOUT 7

tween specific reading abilities of persons whose general ability,
if such there is, appears limited.

In investigating the qualities of a book which make it read-
able for adults of limited ability, we have been forced within the
restrictions of a single report to exclude many that are un-
doubtedly quite as important as the one we have studied
difficulty. To identify some of the other qualities, to classify
them into related categories for the purpose of viewing them
analytically and synthetically, and to attempt to indicate the
type of investigations needed to show their relative contribu-
tion to readability are as much as we have been able to do*
This we have attempted to do in chapter iii.

Furthermore, we have studied difficulty of reading material
only as it is related to structural elements used in the expression
of the content; that is, to length and structure of words and
sentences, to number and hardness of different words, and so on.
Emphasis on these elements is not intended to give them a rank
of first importance in determining difficulty. On the contrary,
it seems altogether probable that whether an idea is abstract or
concrete, whether it is familiar or unfamiliar, are more impor-
tant issues in determining difficulty than whether that idea is
expressed in words of one or several syllables.

Why, then, have structural elements been given precedence
in this investigation? The answer lies frankly in the fact that
they lend themselves most readily to quantitative enumeration
and statistical treatment. Within certain admitted limitations,
subjective opinion cannot gainsay the evidence they present.
If, therefore, we can show that structural elements bear signifi-
cant relationship to difficulty, we not only shall have proof that
reading materials can be made more readable by attention to
form of expression, but we shall have reason to isolate other less
tangible elements and attempt to resolve them into objective
terms for further investigation.

In the hope that the results of the study may be useful in pre-
paring and selecting readable books for the largest possible
audience, we have devoted our attention to the difficulty of

8 WHAT MAKES A BOOK READABLE

what is termed general reading materials. They are prepared for
no specialized group. They aim to present no narrow interests.
They are designed for the general reader whose reading is carried
on without regard for vocational or professional interests. Such
reading matter characterizes newspapers, general magazines,
and general books of fiction and non-fiction. So varied is the
content of this type of material, however, that one is forced to
question whether it can truly be termed general reading or
whether there actually is any general reader. It may be argued
that this material, despite the fact that it is non-technical, rep-
resents fairly specific interests of several kinds of readers. Simi-
larly, persons co-operating in the study were of many different
sorts, racially and socially. It is impossible to say without fur-
ther investigation how much these limitations affect our identi-
fication of elements of difficulty. Perhaps structural elements
are not generally related to difficulty. It may be that their rela-
tionship to difficulty is indeed highly particularized, varying
with the reader and with the nature of the content read. Far
more extended study is needed before this issue can be settled.

Then, again, the elements of difficulty which we have identi-
fied in this study operate only when reading is done for the
single purpose here defined to obtain a general impression of
what is read in the form of a summary statement. But there are
other purposes in reading: to follow the plot of a story, to gather
specific details, to evaluate the worth of an expressed opinion,
to secure emotional enjoyment, to determine the motive of an
author, to support an argument, and to obtain a large number
of other outcomes. All of these are adult purposes for reading,
the relative importance of which has not been discovered* In
the present study we have assumed that adults read most often
to get "the gist" of the content, a general notion of what is
read. It is to this kind of understanding that the structural
elements reported here are related, and it is with respect to this
kind of understanding that they are termed elements of diffi-
culty. According to our findings, the number of different words
in a selection, the number of prepositional phrases contained in

WHAT THE REPORT IS ABOUT 9

the selection, and the proportionate occurrence of polysyllables
bear a significant relationship to the difficulty a reader experi-
ences in reading to obtain a general impression of the content.
The question may well be raised as to whether they would bear
the same relationship to difficulty were the reading done for any
other purpose.

If elements of difficulty as we have defined them should be
found by extended investigation to bear a similar relationship
to difficulty, regardless of the purpose for which one reads, then
to that extent they might be termed "general" elements of diffi-
culty. If, on the other hand, they should be found to vary in
their relationship to the desired outcome, then the definition of
elements of difficulty would of necessity be as particularized as
the outcomes themselves.

Finally, the use of identified elements in predicting the diffi-
culty of specific reading materials for limited readers is also
restricted. For example, in estimating the difficulty of Roose-
velt's Looking Forward, we have done no more than predict the
difficulty it possesses structurally for readers of limited ability
who read it to obtain a general impression of the content.

What we have attempted to show thus far is that the findings
of this report are not applicable beyond the realm from which
they have been derived. The findings with respect to readability
pertain to but one aspect, difficulty. This aspect is further re-
stricted to a study of structural elements in reading material re-
lated to difficulty when reading is done for the purpose of ob-
taining a general impression of what is read. The materials
studied represent a variety of subjects and presumably a va-
riety of reading interests. The individuals used in the investiga-
tion are homogeneous only with respect to reading ability, which
is generally limited. The classification of materials as "easy" or
"difficult" for readers of limited ability is based solely on struc-
tural elements without regard for such qualifying factors as in-
terestingness, familiarity of content, or purpose of reading. It is
with all of these qualifications in mind that we present this ini-
tial approach to the problem of "What Makes a Book Read-

io WHAT MAKES A BOOK READABLE

able." Whatever interpretations are made of the findings pre-
sented throughout the report must be in keeping with the quali-
fications already stated. Interpretations beyond these limita-
tions are wholly unjustified.

In order to give a brief picture of the study as reported in this
volume, the remainder of the chapter is devoted, first, to a sum-
mary of the steps of procedure followed and the findings ob-
tained in that part of the investigation which is presented in
the first six chapters; second, to a survey of the practical appli-
cations of the findings suggested in chapters vii and viii; and,
finally, to a forward look at the nature of future investigations
toward which this report aims to point the way.

STEPS OF PROCEDURE

i. The first step in the study was to secure a list of qualities
of a book which may contribute to its readability for adults of
limited education. In this connection we made a survey of cur-
rent literature for the purpose of discovering what is being writ-
ten about readability and what meaning is being attached to
the term. It was soon apparent that whereas writers and in-
vestigators occasionally mention readable books, they rarely
indicate clearly what is implied by the word readable.

Because librarians, publishers, and teachers and directors of
adult classes manifest growing interest in the question of what
makes a book readable for a particular reader, we sent letters
of inquiry to a large number relative to the factors which they
believe influence readability. A detailed account of the pro-
cedure followed is presented in chapter ii. A total of 288 sug-
gested factors was compiled from approximately 100 responses
to the inquiry. They were classified into twenty-four general
aspects under four major categories: format \ general features of
organization, style of expression and presentation, and content.
The classified list appears in Appendix A. To supplement this
list, reactions were also secured from 170 library patrons who
came to readers' advisers for guidance. These reactions were
stated in terms of factors which they believed made a book read-
able or non-readable.

WHAT THE REPORT IS ABOUT 11

The findings thus obtained were not intended to solve the
problem of what makes a book readable. Their use was confined
to a preliminary definition of a series of problems about which
facts are needed before a book can be labeled "readable" or
"unreadable" for a particular reader. One aspect of readability
proposed by the findings of this survey was isolated for inten-
sive study. This relates to ease or difficulty of reading material,
as determined by the presence of certain structural elements of
written expression that are related to difficulty.

2. Before determining these elements in general reading ma-
terials, it was necessary to discover how well adults read such
materials. Two series of tests were devised from books, maga-
zines, and newspapers. One contained passages of fiction; the
other, non-fiction of a general informational sort.

In giving the tests, an attempt was made to include as many
levels of reading ability as possible. Use was accordingly made
not only of adults who were attending school and were classified
at different educational levels but also of relatively heterogene-
ous non-school groups. It was hoped that the groups would be
sufficiently diversified with respect to abilities, interests, and
educational background to give a fair representation of the gen-
eral reading public and at the same time an adequate sampling
of readers of limited ability. A description of the groups tested
is given in chapter iii.

The average reading score made on each item by all adults
tested was interpreted as the criterion of difficulty for that item.
For example, if the average score on one item was higher than
the average score on another, the first was assumed to be easier
than the second. On the other hand, if the score on the first was
lower than the score on the second, a contrary assumption was
made. Further information concerning how well adults read was
obtained through the use of standardized oral and silent read-
ing tests which interpret reading achievement in terms of grade
norms.

3. Ample evidence was found in the average reading scores of
persons tested on the passages from books, magazines, and
newspapers to indicate that reading materials represent varying

i a WHAT MAKES A BOOK READABLE

degrees of difficulty. Our next step, therefore, was to discover
what elements in those materials influence difficulty. We ac-
cordingly analyzed each item for variants in expression which
might be related to ease or difficulty. Some variants pertained
to vocabulary to its range, frequency of usage, and familiarity.
Others related to the sentence to its length, structure, and use.
A number of others concerned paragraph development and or-
ganization. In all, more than eighty variants of expression were
discovered in the items. Of these, sixty-four were found open to
quantitative enumeration and were therefore retained for study.
They are described in chapter iv.

With this information, it was possible to compare by the meth-
od of correlation the difficulty of the test items with each of sev-
eral elements characterizing their content. This method iden-
tifies elements of difficulty in terms of the relationship existing
between the occurrence of the expressional variants in the items
and the criterion of difficulty, that is, the average reading score
made on the items by persons taking the tests.

In a similar way, elements of difficulty were identified for par-
ticular groups of readers best readers and poorest readers whose
classification as such was determined by their reactions to the
test, after the manner described in chapter iv. The average
reading scores of these groups were taken as criteria of the diffi-
culty which they encountered in reading the test items and were
correlated with the occurrence of expressional variants in the
items as before. The size of the coefficient of correlation thus
obtained indicates the degree of relationship which a particular
variant in expression bears to difficulty. For example, a coeffi-
cient of .520 for percentage of easy words shows that this ele-
ment has a closer relationship to difficulty than does percent-
age of simple sentences with a coefficient of .180.

The direction of the association between any particular ele-
ment and difficulty is designated by the sign of the coefficient.
The elements just cited correlate positively with ease. In other
words, a high percentage of easy words and of simple sentences
may be taken as an index of easy reading material. A low per-

WHAT THE REPORT IS ABOUT 13

centage of the same elements indicates difficult material. A co-
efficient of .380 for percentage of polysyllables, on the other
hand, indicates a relationship negative for ease and positive for
difficulty. Hence, the greater the number of polysyllables in a
selection, the greater the degree of difficulty inherent in it.

4. The identification of elements of difficulty in general read-
ing material gave rise to the question of how to use these ele-
ments in a more reliable estimate of the difficulty of similar ma-
terials. It was believed that the answer would furnish librarians
and readers' advisers with a scientific technique for determining
what materials are of appropriate structural difficulty for read-
ers of known ability. Furthermore, it would suggest an objec-
tive means whereby writers and publishers can determine
whether a particular mode of expression will offer a serious ob-
stacle to adults of limited reading ability.

Two techniques were devised for estimating the difficulty of
general reading material by the use of the significant elements.
These techniques are presented in chapter iv. Their application
to specific books, magazines, and newspapers is illustrated in
chapters v and vi.

THE MAJOR FINDINGS

The most important findings revealed by this study may be
summarized as follows:

i. Measurement of the reading ability of 1,690 adults showed
a wide variation in achievement, ranging from a grade equiva-
lent below 2.95 to one above 16.95. About one-sixth of the
adults tested were found to read with a proficiency normally at-
tained by high-school graduates. Approximately the same num-
ber had attained a reading proficiency commonly associated
with the lower elementary grades. Between these two extremes
range the majority of adults tested.

In all probability, explanation for the low reading achieve-
ment reported here lies partly in the selection of cases to which
reference has been made on page n. We were interested not
only in discovering how well adults read in general, but in de-

i 4 WHAT MAKES A BOOK READABLE

termining how many persons in adult classes, in organized adult
groups, and in representative communities have attained some
degree of reading facility, and yet are unable to engaged in ma-
ture reading activities. The fact that many adults cannot read
understandingly a large proportion of available materials justi-
fies a concentration of effort on the problem of ascertaining
what makes a book readable for them.

i. Librarians, publishers, and others interested in adult edu-
cation are in notable agreement with respect to the factors
which in their opinion influence the readability of a book for
readers of limited ability. All three groups believe that factors
of content are of greatest importance. They rate factors of style
of second importance; factors of format, third; and general
features of organization of least importance. In other words,
they believe that a readable book, first of all, must contain con-
tent relating to the reader's interest. For example, Fleming's
Brazilian Adventure may be very readable for one person be-
cause he is interested in travel or because he is drawn as by
magic to thrill-packed adventures in the jungle. Another may
find the same book quite unreadable. He, too, wants a travel
story, but of another sort. He prefers to journey happily and
informally over well-marked highways, as in Winn's The Mac-
adam Trail; Ten Thousand Miles by Motor Coach.

In the second place, according to a majority of the judges,
the readable book has a pleasing style. It tells the reader "what
he wants to read about" in a manner that makes him look ahead
to what is yet to come with anticipation and look back over
what has gone before with satisfaction. It neither vexes him
with overcomplexity nor with oversimplicity. He can react nat-
urally and favorably to the material because its style of presen-
tation fits his needs and tastes.

The judges are fairly well agreed, also, that an attractive
format is of some importance to readability. They believe that
the most ambitious reader may be daunted by a 6oo-page book,
even though the content is what he wants to read about and
the style agreeable to him. The fact is, he is not that much in-

WHAT THE REPORT IS ABOUT 15

terested, when all he truly wants to know could be told in a book
one-fourth as long. On the other hand, a book may be too con-
densed and too brief to be readable. Or, again, the illustrations
may be too few or too many, too detailed or not detailed enough,
too garish or too dull, to satisfy the reader.

Finally, according to the judges, how a book is organized has
some bearing on its readability. The organization must make
it possible for the reader to get what he wants as easily and
quickly as circumstances allow. If it does not, then the book
falls just short of being readable.

Although considerable agreement was found among the
judges concerning what general qualities make a book readable
for readers of limited ability, there was marked diversity of
opinion with respect to the individual factors that promote in-
teresting content, pleasing style, attractive format, and appro-
priate organization. It is this diversity of opinion, shown in
later sections of this report, that points to the need of more ob-
jective and reliable evidence concerning the qualities of reada-
bility.

3. That ease or difficulty of a book is a potent factor in effect-
ing readability for persons of limited education is generally con-
ceded by the readers themselves. They report a certain book
readable because "it is easy to understand"; "it has no big
words in it"; "it is written so you can read right along." Such
testimonies have led us to inquire, What are the elements in a
book that make it "easy to understand" and allow the reader
"to read right along"? The major part of the present study is
devoted to finding an answer to this question.

4. The facts obtained relative to difficulty show that forty-
four structural elements of the kind mentioned on page 12
bear some relationship to difficulty. Sometimes the relationship
is in the direction of ease, as in the case of simple sentences,
personal pronouns, monosyllables, familiar words, and so on.
Again, it is in the direction of difficulty, as for long sentences, a
widely diversified vocabulary, hard words, and certain other ele-
ments that exert a negative influence on ease of comprehension.

1 6 WHAT MAKES A BOOK READABLE

More than twenty elements were found to bear significant rela-
tionship to difficulty. Among them are such elements as ex-
plicit sentences, length of sentence, simple sentences, and words
not known to 90 per cent of sixth-grade pupils. Elements bear-
ing little relationship to difficulty include parenthetical expres-
sions, words beginning with w, /*, or <?, bisyllables, and others.
Their influence on difficulty is notably less than that of the
most significant elements.

The findings indicate that the relationship of certain elements
to difficulty varies also with different kinds of readers. The less
able the reader the higher the relationship. For example, the
relationship between figures of speech and difficulty is markedly
high for readers of limited ability and markedly low for good
readers. Hence, the presence of that element cannot be taken
as a reliable indicator of the difficulty of the selection in which
it occurs, save for a particular kind of reader.

5. The findings show that it is possible to estimate the diffi-
culty of reading materials by the use of any one significant ele-
ment of difficulty. They show also that a more reliable estimate
can be made by the use of several elements. For instance, count-
ing the number of different words in a selection gives a fairly
good measure of its difficulty. Counting other elements prepo-
sitional phrases, simple sentences, and personal pronouns
gives a much better indication of difficulty. As determined by
the procedures adopted in this study, the best estimate of the
difficulty of a selection involves the use of eight elements : num-
ber of different hard words, number of easy words, percentage
of monosyllables, number of personal pronouns, average sen-
tence-length in words, percentage of different words, number of
prepositional phrases, and percentage of simple sentences. That
smaller combinations of the same elements give about as good
an estimate of difficulty at an expenditure of considerably less
time and effort is shown in a later chapter.

6. Interesting findings resulted from an analysis and classifi-
cation of 350 books according to structural difficulty. Their
predicted scores distribute themselves in a close approximation

WHAT THE REPORT IS ABOUT 17

of the normal curve. Relatively few books were found to be
simple enough to rank among materials suited to a reading
achievement lower than fifth grade. The largest number rank
at an area of difficulty termed "average." They present no
greater difficulty from a structural point of view than school
readers prepared for sixth grade and junior high school. Some
few books among the 350 rank "difficult" or 'Very difficult/'
indicating that they afford structural obstacles beyond the com-
prehension of most adults of limited reading ability.

Each of these findings is based on a quantity of supporting
data, tabulated and summarized in the remaining chapters of
the report and in the Appendix. Statistical proof for the accu-
racy and reliability of the findings, their interpretation for prac-
tical purposes, and recommendations for supplementing them
by further investigations are also presented in later sections of
this volume.

APPLICATIONS OF THE FINDINGS

The problem undertaken in this study was conceived as one
of practical value in improving the reading habits of adults.
Ample testimony has accumulated to indicate that obstacles to
easy, enjoyable reading add to the already complex task of pro-
viding for the varied needs and interests of the reading public.
Hence, the identification of difficulty-elements which are in-
herent in reading materials promises a means whereby those
now available may be better adapted to adult readers and new
materials be prepared which the largest possible audience will
find readable. As other aspects of readability are studied, it
should be possible to accomplish both undertakings with still
greater success.

How information concerning factors of readability and ele-
ments of difficulty may be put to practical use is considered in
chapters v to viii. In chapters v and vi we have illustrated how
the findings of this study may be used by librarians and others
in estimating in an objective and reasonably reliable manner
both the relative and the absolute difficulty of general reading

1 8 WHAT MAKES A BOOK READABLE

materials. If our ranking of these materials approximates the
order of difficulty that librarians would assign to them on the
basis of opinion alone, then the study has been worth while in
giving objective support to subjective practice. If, on the other
hand, the difficulty assigned does not accord with the judgment
of librarians, then the worth of the study lies in part in propos-
ing an objective method of estimating reading difficulty and in
part in suggesting problems which merit further consideration,

In chapter vii we have recommended for librarians, teachers,
and advisers of adult reading certain procedures for estimating
how well a person can read and what reading material is of ap-
propriate difficulty for him. Some of these procedures are al-
ready in use. Others have been tested experimentally. Still
others are purely theoretical. Their usefulness in a practical
advisory situation remains to be shown by librarians and others.
All of the proposed procedures aim to do the same thing, name-
ly, to translate the ability of the reader and the difficulty of
reading material into common terms so that a knowledge of the
one will supply knowledge of the other. Information concerning
the reader's ability undoubtedly can be secured best by li-
brariansj readers' advisers, and teachers, as has been done in
the past. Facts concerning the difficulty of reading material
probably should be supplied by some centralized agency whose
chief function would be to compute the index of difficulty for
general reading material and to make these indexes known
among publishers, librarians, and counselors of adult reading.

If we are to accept the evidence that much of the general
reading material now available is too difficult for the adult of
limited reading experience, and if we can prove that a simplifi-
cation of certain elements reduces the difficulty of the material,
then the preparation of simple books may be greatly en-
couraged. But the simplification cannot be left to sheer guess-
work. Among other things, it must take cognizance of reliable
evidence concerning the modes of expression that please or dis-
concert particular classes of readers. In chapter viii we have
suggested tentative standards of difficulty to serve as guides in

WHAT THE REPORT IS ABOUT 19

preparing material for adult readers at given levels of ability.
Inasmuch as these standards pertain only to structural ele-
ments of expression, they should be supplemented by others as
rapidly as they are identified.

It is not expected that an author can revolutionize his style
of expression to fit these standards all at once. In most cases it
is not desirable that he should do so. Conscious attention to
such matters may prove disastrous. What any writer can do,
however, is to become familiar with materials that meet certain
standards and then consciously aim to adapt his manner of
writing to the level of the greatest possible audience.

SUGGESTIONS FOR FURTHER STUDY

It seems important to repeat that the present report is not
intended to settle the question of what makes a book readable.
On the contrary, its major purpose is to open up the whole ques-
tion and reveal some of its manifold ramifications. By so doing
at least three outcomes are hoped for: first, that it will indicate
the possibility of identifying objectively some of the charac-
teristics which determine readability for particular classes of
readers; second, that it will suggest the possibility of controlled
investigation of these various characteristics now in urgent need
of consideration; and, third, that it will pave the way to the
development of objective procedures for selecting and prepar-
ing readable books to supersede impressionistic judgment.

That the present study of difficulty is not a model to be fol-
lowed by investigators of other aspects of readability is admit-
ted by the many limitations presented earlier in the chapter.
A series of controlled investigations covering all the major as-
pects of the problem is needed to remove these limitations and
to exemplify methods of procedure that will produce the most
reliable findings. We propose in this connection that questions
of the following sort be considered:

1. To what extent does difficulty of content affect readability for different
kinds of readers ?

2. What other qualities of content than structural elements are related to
difficulty of understanding?

ao WHAT MAKES A BOOK READABLE

3. How are elements of expression related to difficulty in different kinds of
content: history, travel, science, etc.?

4. To what extent is a reader's interest in a selection related to his ability to
read the selection satisfactorily, regardless of its structural elements?

5. Does particular reading material present equal difficulty when read for
different purposes?

6. Is the difficulty of a selection related to a reader's impression that the se-
lection is easy or hard?

7. To what extent does a reader's familiarity with a subject affect the in-
fluence of structural elements on understanding ?

8. Are structural elements related to difficulty in the same degree for readers
with different racial and cultural backgrounds?

9. How much weight can be given to structural elements in determining the
total difficulty of a selection ?

It is important that each of these and other problems sug-
gested by this report be subjected to carefully controlled inves-
tigation in order that the facts concerning difficulty be made as
reliable and comprehensive as possible. When the evidence has
been extended far enough to define difficulty for different kinds
of readers, reading different types of material for different pur-
poses, it will be possible to carry experimentation in writing
much farther than this report suggests. By utilizing elements of
difficulty in varying amounts, we may eventually determine the
appropriateness of the material for particular groups of readers
in a variety of reading situations.

It is also important that such a series of studies be supple-
mented by further investigation and experimentation of a simi-
lar nature with respect to other aspects of readability. With the
combined findings of all these efforts to serve as a guide, more
reading materials should be produced for adults of limited read-
ing ability, to the end that their present reading ventures will
develop ultimately into permanent reading habits.

CHAPTER II
WHAT IS A READABLE BOOK?

A ALREADY mentioned, recent investigations indicate
that the quality of reading material which is called
"readability" influences directly the reading habits of
adults. It seems equally true that accessibility of books pro-
motes wide reading. An essential step, therefore, in improving
reading habits is to provide adults with materials that will be
readable for them.

But what is meant by the term readable^ Does it imply only
qualities which are inherent in a book? Is it dependent upon
individual characteristics of a particular reader? Or does it ex-
press a certain relationship between qualities inherent in the
book and individual characteristics of the reader? These ques-
tions indicate the complexity of the task of selecting and pro-
ducing readable books for different kinds of readers. One of the
first steps in a comprehensive study of the problem is to deter-
mine the qualities of a book which may influence its readability,

We have accordingly undertaken, within certain limitations,
to discover the meaning librarians, publishers, and others in-
terested in adult education attach to the term "readability."
If a book is a composite of several qualities, how much impor-
tance do these persons assign to the various components? It is
not our intention to arrive at a valid definition of readability
from sheer opinion. It is rather to determine the extent to which
readability has the same or different meaning for persons en-
gaged in preparing and selecting readable books for others to
read.

We have hoped to accomplish at least three other objectives:
first, to secure a comprehensive list of the qualities of a book
which may contribute to its readability; second, to identify one
aspect of readability for more intensive study at this time; and,

22 WHAT MAKES A BOOK READABLE

finally, to suggest the general character of subsequent studies
which are needed before a particular book can be designated
readable or non-readable for a particular kind of reader.

CHARACTERISTICS OF A READABLE BOOK AS
DESCRIBED IN RECENT LITERATURE

An examination of the literature bearing on adult reading
shows that an interpretation of the term "readable" occurs with
less frequency than does the term itself. A few writers have at-
tempted to define readable materials in terms of a specific kind
of reader on the basis of observation or investigation. Others
have implied a broad meaning of readability. The majority have
referred to readable books but have failed to indicate the mean-
ing which they attach to the word readable.

In the language of the dictionary, a readable book is one
"that may be read with satisfaction or interest; that is attrac-
tive in style or treatment; that is easy and pleasant to read."
There is implied in this definition the idea that the test of a
readable book lies in the pleasurable reaction which it creates
in the reader by its content, by its attractive style, or by the
ease with which it can be read. In the light of this definition,
therefore, we need to discover three things : what is interesting
to different groups of readers, what style is attractive to them,
and what material is easy for them to read. With this infor-
mation we can then select for a particular group of readers a
book that in the terms of the dictionary will be readable for
them.

Knowledge is readable, according to James Harvey Robin-
son, "when it is humanized." 1 In The Humanizing of Knowl-
edge he holds that a book for the general reader "with no great
surplus of time, preparation, attention or initial interest" must
do three things: first, it must enlist the reader's attention; sec-
ond, the facts and information must be presented in terms and
in an order which will be understood by him and will fit into his

* James Harvey Robinson, The Humanizing of Knowledge. New York: George H.
Doran Co., 1923. Pp. 120.

WHAT IS A READABLE BOOK? 23

way of looking at things; and, finally, the significance of the in-
formation in its bearing on the reader's thought and conduct
and his judgment of others should be wisely suggested. 2

How can these results be attained? When does a book enlist
the reader's attention? How can facts and information be pre-
sented so that they will be understood? In part, according to
Robinson, by means of a subject that is vividly and persistently
interesting; in part, through use of the story form, which makes
the best and surest kind of appeal; in part, through "good little
books, easy to slip into one's pocket or bag"; and, finally,
through simplicity of language and style, which is lacking alto-
gether too frequently.

He says,

Most books are simply too long and too hard for even ambitious and in-
telligent readers. For to be simple is to be sympathetic and to endeavor to

bring what one says or writes close up to those one is addressing And

the great art in writing is not to exhibit one's own insight and learning but
really to influence those whom one is aiming to influence. 3

The Subcommittee on Readable Books of the American Li-
brary Association has defined a readable book for the middle
group of readers ranking between the specialist and the person
of extremely academic turn of mind at one end of the scale and
the reader of light fiction at the other. For this middle group,
the Committee designates a readable book as one having seven
characteristics: simplicity of knowledge, non-technical treat-
ment, brevity of statement, fluency, adult approach, vitality,
and certain physical features. 4 Although the Committee has
made some concessions on several points, it stands firm on two
which it believes essential to readability. The book must be
simple enough to be understood, and it must have a degree of
vitality; otherwise it is not readable. How simple a book must
be to be readable for different types of readers and how vitality
can be measured are problems suggested by the Committee's
point of view.

*Ibut. 9 p. 105. * Ibid., pp. 88-89.

* Emma Felsenthal, Readable Books on Many Subjects (Chicago: American Library
Association, 1929), p. 4.

24 WHAT MAKES A BOOK READABLE

Waples has reached the conclusion from his study of what
people actually read that librarians and publishers must find
out what the reader wants to read about and what are his pref-
erences concerning author, style, length of treatment, and the
like. 5 In this way they may increase the reader's satisfaction
with what he reads.

Chancellor's inquiry concerning the suitability of available
reading material for native-born adult illiterates and near-
illiterates has brought together the opinions of educational
workers and librarians who have had experience with the prob-
lem either directly or indirectly. 6 The net result of his inquiry
shows a general opinion that much of the published material
now available for the groups in question fails to be readable for
one of three reasons. It is too difficult; it progresses too rapidly
in difficulty; or its content is not sufficiently interesting to
adults. Since this opinion is based on experience with illiterates
and near-illiterates, it is presumably reliable. It suggests the
need for objective investigation to determine what elements in
materials contribute to difficulty for such readers; how rapidly
a story can progress before it becomes disconcerting; and what
content is most interesting. Knowing the answers to these ques-
tions, we can select books which will be more readable for these
readers, if such books exist; or we can prepare materials better
suited to their needs and tastes.

A recent study by Conrad is concerned with the appropriate-
ness of elementary school reading texts for use in teaching adult
illiterates and near-illiterates. 7 The findings show that three fac-
tors influence vitally the choice of books the quality of vocab-
bulary, repetition of words, and good sentence structure. Three

s Douglas Waples, "The Relation of Subject Interests to Actual Reading," Library
Quarterly \ II (January, 1931), 42-70.

6 John Chancellor, "Available Reading Material for Native-born Adult Illiterates
and Near-Illiterates." Mimeographed. Washington, D.C.: Bureau of Prisons, Depart-
ment of Justice, 1933. Pp. 35,

? L. R. Conrad, "Investigation in Reading Material for Native Adult Illiterates and
Near-Illiterates." Mimeographed. Chillicothe, Ohio: United States Industrial Reform-
atory, 1933.

WHAT IS A READABLE BOOK? 25

other factors hygienic requirements, nature of the content,
and interest factors are of slightly less importance. Conrad
has gone so far as to describe what seem to be the requisites of
reading material appropriate for the men in his prison-school.
It is significant that all the requisites are aimed to make the
material easy to read, implying that for this group material
which can be read is readable.

The foregoing discussion indicates the variety of meanings
which writers attach to the term "readable/* Some are based
on sheer opinions; others, on observation and experience; and
others, on experimentation. It seems important to note that
despite the variety of interpretations offered by different writ-
ers, two aspects are generally held essential to a readable book.
One is interesting content, and the other, a presentation simple
enough to be understood. The extent to which these two as-
pects are commonly considered important by librarians and
others who advise adults in their reading, by publishers, or by
persons interested in adult education, will be shown in subse-
quent sections of this report.

COMPILING A LIST OF POSSIBLE FACTORS THAT MAY

INFLUENCE READABILITY FOR READERS OF

LIMITED ABILITY

According to the plan outlined in this chapter, we have under-
taken to discover the trend of opinion relative to the meaning
of readability for the purpose of defining lines of investigation
needed in the field. In order that all who participated in the
study might have opportunity to consider the same factors, a
list was prepared which included as many potential factors of
readability as could be obtained. The list was compiled from
responses of a large number of librarians, readers' advisers,
publishers, and other persons interested in adult education who
were questioned as to the factors which they believe contribute
to readability for adults of limited education. It was desirable
that their replies to the inquiry should not be influenced by sug-
gestion. The term "readability," therefore, was not defined. In-

a6 WHAT MAKES A BOOK READABLE

stead, each person was asked to interpret the term for himself
and then to define or explain quite explicitly any factor he
listed. Approximately one hundred replies were received to the
inquiry. A few librarians good-naturedly admitted some reluc-
tance about "going on record" concerning factors which make
books readable, despite the fact that they are daily engaged in
passing judgment on the readability of particular books.

When factors listed in the various replies were examined, it
was found that they could be classified into a single list. Certain
ones were grouped as components of a larger aspect of reada-
bility, and these, in turn, as components of a still larger aspect.
For example, suggested factors included "numerous illustra-
tions," "cartoon- type of pictures," "illustrations adjacent to
the text," "appropriate diagrams/' "colored inserts," and so on.
These factors obviously belong to a single aspect, illustrations,
which is a subdivision of format.

The classified list of 289 factors as finally compiled contained
four major categories designated by Roman numerals: (I) For-
mat or Mechanical Features, (II) General Features of Organiza-
tion, (III) Style of Expression and Presentation, and (IV) Con-
tent. Under each category were placed a number of related gen-
eral aspects designated by Arabic numerals* Under each of these
were listed the specific factors, a y b> c, and so on. A section of
the classified list is shown on p. 27.

Since direct contradiction occurred between certain factors
and a degree of overlapping among others, the composite list
was by no means free from inconsistencies. For example, one
person held that a readable book for a reader of limited ability
should employ questions and answers; another stated that it
should use no questions and answers; whereas a third believed
that a judicious use of questions and answers promotes reada-
bility. All three opinions appear in the composite list. Since no
factor was tabulated unless it had been suggested in response to
our inquiry, the list was obviously not exhaustive.

A SECTION FROM THE LIST OF POSSIBLE FACTORS
OF READABILITY IN BOOKS

I. FORMAT or MECHANICAL PTSATTJRES

j 1. Size of Book

a. Small

b* Avernse

c. Larger than a textbook

d. About 5" by 8"

? ^out 14 cm* by 16 ciiu

. HKnt-weisht

a; Coiaf of table

li i.'ot forbicldina

1, Abo^i t 2Q onu oy 14 cm

1 2. Number of Pages

a. Brief

b. About 50 papces

c. About 75 pages

d. About 64-96 oaffes

e. About 300 Dagos

f. About 125-5L5O pr>

. About 200 pa^es

h. About 30O-400 PXD.

! About 200-3OO DP

1 3, -Duality of Paper

a. Opaque

b. Dull-surfaced

f. Even-colored

aiite

e. Not white

f * Pleasant to. touch

K . Qood

h. Glossy

1 4. Kind of Type and Printing

a. Larp;e

b. Good-sized

c. Medium

d* Small

e. About 11 pt

About 12-14 pt^

f;* Not under 8 pt

tu No* 7. Old Style

1* 4 pts leading

k. Well spaced

1* Open face

ITU Black ink

u Dull lak

28 WHAT MAKES A BOOK READABLE

TECHNIQUE OF OBTAINING OPINION RELATIVE TO
FACTORS OF READABILITY

In order to determine the relative importance attached to the
factors of readability, the classified list was sent to the persons-
who had answered the original inquiry and to a number of addi-
tional librarians and publishers. They were asked to evaluate
each major category and each general aspect for its influence on
readability, and to indicate what they believe to be the impor-
tance of specific factors for readers of limited ability. The fol-
lowing directions were given :

METHOD OF SCORING AND CHECKING

Three columns are set up at the right of each page of listed factors, desig-
nated A, B, C. Column A is to be used for checking specific factors; Column
B, for evaluating general aspects; and Column C, for evaluating the four
major categories.

Examine the complete list of factors and note the general aspects and cate-
gories into which they have been classified, in order to familiarize yourself
with the general set-up of the list. You will note that there is direct contra-
diction among some factors, apparent overlapping of others, and close simi-
larity among others. This has resulted from including all possible factors of
readability suggested by the correspondents.

First step. Look over the factors, designated a, , c, etc., and decide which
ones, in your judgment, make for readability. Indicate your decision by check-
ing such factors (i/) in Column A. If you wish to show that certain factors
are of special significance, indicate by (i/i/)- Leave blank spaces after fac-
tors that you believe are insignificant or do not make for readability. Space
has been left for additional factors. Please include any that you think have
been omitted.

Second step. Consider the total value of the general aspects i, 2, 3, etc.,
in each major category as equal to 100 points. Look over the general aspects
in Category I, and evaluate their relative importance in promoting reada-
bility. Distribute the total value, 100 points, among these subdivisions to
indicate their proportionate values. Write the values in Column B, opposite
each. Be sure that the sum of all values assigned to aspects I, 2, 3, etc.,
equals 100 points, which is the total value of Category I.

Next, do the same thing for Category II, then for Category III, and finally
for Category IV. In each case, distribute 100 points among their respective
general aspects, writing in Column B. Each category should total 100 points.

Third step. Now consider the total value of all four categories, I, II, III,
IV, as equal to 100 points. Decide what proportion of 100 points best repre-
sents the value of each category in influencing readability. Distribute the 100
points among the four categories, as your judgment dictates. Write the as-
signed value in Column C after each category. The sum of the four values
should total 100 points.

WHAT IS A READABLE BOOK? 29

THE NATURE OF THE REPLIES

About ninety persons checked and returned the list, having
followed all or part of the directions. Occasionally they omitted
the first or second step* In every case, however, they carried
out the directions concerning the third step. It is probably
easier to evaluate the relative influence of large categories on
readability than to judge the importance of specific factors.

Some of the judges had acquired definite views concerning
the factors which contribute to readability and checked the list
with considerable assurance. Others approached the list with
extreme conservatism, declaring that there was no item which
could be endorsed or eliminated without some qualification. A
few chose to check only the factors that they believe generally
contribute to readability, since their experience with individual
readers had been limited. Similarly, several publishers checked
those factors which they consider indispensable in the actual
practice of producing readable books for all readers.

A number of judges misunderstood our motive and felt that
we were attempting to obtain merely a general impression about
readability. Hence they objected to checking the list, "since
there can be no generalization concerning factors of readabili-
ty." One publisher stated, " I can't bring myself to the point of
believing that the factors of readability can be stabilized and
labeled as this investigation attempts to do." To this statement
we are in mingled accord and disagreement. In accord because
an inquiry such as this one cannot establish a particular aspect
of a book as a factor of readability, nor does it attempt to do so;
in disagreement, because it seems tenable to assume that when
the reading interests, tastes, capacities, and needs of certain
kinds of readers have been discovered, we shall be able to de-
fine a readable book for them.

ORGANIZATION OF THE DATA

Seventy-nine judges carried out all directions of the inquiry.
Of this number thirty-four are librarians or readers' advisers,
sixteen are publishers, and twenty-nine, directors or teachers of

30 WHAT MAKES A BOOK READABLE

adult classes. Their opinions are recorded in this chapter in the
following order:

First, the relative weight assigned to the four major cate-
gories is summarized in terms of the mean judgment, the stand-
ard deviation of the mean, and the range for all judges as well
as for the three groups of judges. The mean judgment is con-
sidered the "characteristic" weight of each category (within
certain statistical limitations). That the inquiry should reveal
some discrepancies of opinion was expected. The degree of these
discrepancies among the different groups of judges and among
judges within a group is indicated by the standard deviation and
the range.

Judgment concerning the four major categories has been
given first for two reasons. One is to give a broad view of factors
of readability as they are evaluated by the judges; the other, to
indicate the order of presentation to be followed later in report-
ing opinion concerning each category.

After the evaluation of the four major categories, judgment
relative to content is summarized according to the separate
factors believed to contribute to readability and according to
the relative influence of the factors when classified into genera]
aspects of content. Opinions concerning style of expression and
presentation are given next, followed by a record of opinions
pertaining to factors of format. Finally, summarized tabula-
tions are given bearing on features of organization which all
groups of judges consider of some significance for readability.

JUDGMENT CONCERNING THE RELATIVE INFLUENCE OF THE

FOUR MAJOR CATEGORIES OF POSSIBLE

FACTORS OF READABILITY

Table I presents the summary of opinion relative to the four
major categories into which possible factors of readability have
been classified. The same facts are shown graphically in Fig-
ure i.

Factors which all judges believe have greatest influence on
readability belong to content. Its mean value is 33.64 per cent.

WHAT IS A READABLE BOOK? 31

Second in influence are factors of style, with a mean value of
30.71 per cent. According to the combined opinion of all judges,
then, if you give a reader a theme which interests him, that is,
one that he wants to read about, you have the problem of
readability one-third solved. Furthermore, if in addition you

TABLE I

SUMMARY OF JUDGMENT CONCERNING THE RELATIVE INFLUENCE ON READABILITY
OF THE FOUR MAJOR CATEGORIES

MAJOR CATEGORY

ALL PERSONS

LIBRARIANS

PUBLISHERS

OTHERS INTER-
ESTED IN ADULT
EDUCATION

Range

I. Format

20.26
iS-38

30.71
33-64

7.68
7.04

9.17
13.11

45-5
40-3

50-0
75-7

24.13
15.71

32.74
27.42

7.64

6. 59

8-39
9-95

45-7
26-3

SO-20
50-7

17.08

15- 4*

32.92
34.58

5.20
6.37

8.j 7
12.83

25-10
30-5

50-25
50-10

17.92
15.20

27.50
39-37

6,37
7-24

9.24
12-54

30-5
40-5

40-5
75-20

II. General Features of Or-
ganization . . ,

III. Style of Expression and
Presentation

IV. Content

Per cent

Others

General c . .

., features of Style of ex-

Format rt ^ pression and Content

organiza- r

tion Presentation

-Opinion concerning the influence of classified factors on readability

.1* A. *.4>vr1ss. !* -L i-innJ. n U. d n M f* ^ J r* f% A J nn.f*.f* 4.1^ *+ t* * ~. J.'i

find out the style which best fits his needs and tastes, that is, the
scope of vocabulary and the kind of sentences which he reads
easily and the type of presentation he reads with pleasure, then
you have the final solution of the problem close at hand. At least
it is 64.35 P er cent solved, as measured by the combined mathe-
matical judgment of seventy-nine judges.

32 WHAT MAKES A BOOK READABLE

Publishers and others interested in adult education also give
content a place of first importance, as indicated in the table by
their respective mean values of 34.58 per cent and 39.37 per
cent. Publishers, however, give style so nearly the same value
on the average that no significant precedence can be claimed by
content. Directors and teachers of adult education, on the other
hand, generally consider content of greatest worth, one person
giving it a value of 75 per cent, and none less than 20 per cent.
Librarians as a group consider style of slightly greater influence
on readability than content.

All groups agree fairly well that less than 40 per cent of the
total contribution of all factors toward making a book readable
is made by format and organization. It appears on the basis of
this evidence that librarians, publishers, and others would make
readability depend finally upon agreeable content and style.
Nevertheless, they recognize the importance of attractive for-
mat, size, weight, general mechanical set-up, and a desirable
organization. How closely this opinion coincides with the actual
facts remains to be determined by experimentation with dif-
ferent kinds of books among many kinds of readers.

In general, the degree of variability of individual judgment
is about the same for all groups, except with respect to content.
A comparison of the standard deviations (<r) for this category
shows that the mean assigned by librarians is more nearly repre-
sentative of all values than is the mean value of any other group.
That is to say, since 9.95 is smaller than other standard devia-
tions, the mean value of content, 27.42, is a more reliable value
than are the mean values obtained for publishers and persons
interested in adult education. It appears, therefore, that li-
brarians are in closest agreement as to the influence of content
on readability. The least diversity of opinion for all categories
except content is found among publishers. This agreement is to
be expected. The very nature of the publishing enterprise prob-
ably creates in the publisher a more critical attitude toward
a book as a whole than exists among the other groups of
judges.

WHAT IS A READABLE BOOK?

WHAT FACTORS OF CONTENT INFLUENCE READABILITY?

We have just summarized opinion relative to the influence on
readability of factors classified into major categories. We have
shown also that factors of content are considered of greatest im-
portance by the mean evaluation of all judges. But content is
a general quality that may be thought of in many ways, in
terms of a specific theme, in terms of its appeal to the reader,
and so on. It seems important, therefore, to examine the vari-
ous aspects of content as they have been classified in the check-

TABLE II

FACTORS RELATED TO CONTENT RANKED OP GREATEST AND LEAST IMPORTANCE
TO READABILITY BY EACH GROUP OF JUDGES

Factors Ranked in Highest One-
Fourth of 47 Factors

Factors Ranked in Lowest One-
Fourth of 47 Factors

2/2* Timely subject matter

ig Theme people and personalities

im Theme travel and business

i/ Theme romance and action

in Theme one of human interest

2J Interesting subject matter

id Theme adventure

ic Theme history not important
ij Theme not just ideas
ik Theme not theories
is Theme opposed to reality
iq Theme not analysis of human ex-
perience

* Refers to classified item* in Appendix A.

list in order to determine which ones are held generally impor-
tant or unimportant. Since the judges who co-operated in the
inquiry are actively engaged in producing, selecting, or recom-
mending books for adult readers, agreement concerning specific
factors that presumably affect readability is to be expected.

Table II lists the factors of content ranked of greatest and
least importance by each of the three groups of judges, impor-
tance being determined by the number of times the factors are
checked. A factor is considered of greatest importance if it
ranks above the third quartile, that is, in the highest one-fourth,
in a distribution of all factors according to the number of times
each is checked by each group of judges. A factor ranking be-
low the first quartile, that is, in the lowest one-fourth of the

34 WHAT MAKES A BOOK READABLE

same distribution, is considered of least importance. It must be
emphasized that the rank of a particular factor solely on the
basis of number of times it is checked does not define its actual
degree of importance. It rather defines the relative importance
which groups of judges have assigned to it. It is altogether prob-
able that a highly important factor may be given little weight
because in the absence of objective evidence only a few far-
seeing persons in a group have grasped its true significance.
Data presented in Table II, therefore, show no more than the
trend of opinion among all groups of judges with respect to
factors of content which they believe make a book readable for
readers of limited ability. In Tables LXVI, LXVII, and
LXVIII, Appendix B, are listed the factors ranked of greatest
and least importance by librarians, publishers, and others in-
terested in adult education, respectively.

Examination of Table II shows that in the opinion of most
judges the content of a readable book for readers of limited
ability should be timely (20) and interesting (2/), and that its
theme should be of human interest (i#), about people and per-
sonalities (ig-), travel and business (im)> romance and action
(i/), or adventure (i^). According to the opinion of librarians
and other persons interested in adult education, the most im-
portant aspect of content is timeliness, irrespective of the theme
presented. Publishers, on the other hand, give more frequent
mention to the nature of the theme, ranking adventure (id]
first; science and invention (i<?) second; human interest (in)
third; and so on. A timely topic (20) ranks seventh. The ma-
jority of judges in all groups "double-checked" ig and id to
indicate that they believe themes about people and personalities
and about adventure are of special significance in promoting
readability. Librarians have observed profound interest in these
themes among readers of limited ability. Accordingly, some
have endorsed the simplification of popular novels of such
authors as Zane Grey and Oliver Curwood, for the near-illiterate
reader. 8

* Chancellor, op. /., p, 5.

WHAT IS A READABLE BOOK?

Is it important that the content of a book for limited readers
be "not just ideas'* (y), "not theory" (i), that it be "opposed
to reality" (u), or that it avoid "analyzing human experience"
(i#) ? Some individual judges expressed such a belief in our pre-
liminary survey. However, when they evaluated the factors lat-
er they generally considered them of little importance. In other
words, a writer need not avoid theories or realities when writing
for limited readers provided that he is concerned with what is
timely and of human interest. Although the observations of li-
brarians with respect to what people want to read are helpful,
more scientific data are available. Objective evidence concern-
ing the common interests of groups of readers, compiled by
Waples and Tyler, answers the perplexing question concerning
what people want to read about. 9 This evidence undoubtedly
solves one of the major problems of readability.

THE RELATIVE IMPORTANCE OF GENERAL
ASPECTS OF CONTENT

Table III summarizes the value attached to general aspects
of content after the manner described on page 30. The mean

TABLE III

SUMMARY or JUDGMENT CONCERNING THE RELATIVE INFLUENCE ON READABILITY
OF GENERAL ASPECTS OF CONTENT

GENERAL ASPECT

ALL PERSONS

LIBRARIANS

PUBLISHERS

OTHERS INTER-
ESTED IN ADULT
EDUCATION

Range

i. Theme P ,

47.36
37-24
15.40

xi,oa
9.31
8-79

80-10
75~i S
35-

46.77
37-44
15.81

9.16

75-30
50-25
35-o

47-aS
34-43
18.33

IO.20
6.60
8.98

67-30
50-30
30-10

46.51
40.20
13.28

13.23

12.01
8.06

80-10
75-1$
25-0

2. Nature of subject matter

values (Fig, 2) show an unmistakable similarity of opinion
among the three groups of judges. Theme is assigned highest
value by all groups; nature of subject matter, second highest;

9 Douglas Waples and Ralph W. Tyler, What People Want to Read About. Chicago:
University of Chicago Press, 1931. Pp. 31 a.

WHAT MAKES A BOOK READABLE

and unity of content, lowest value. These ratings are in har-
mony with the tabulations presented in the previous section.

It is apparent from the size of the standard deviations that
closest agreement of opinion persists among librarians, and least
agreement among persons interested in adult education. Mean
opinions of the latter group regarding theme and nature of sub-
ject matter are considerably less reliable than the opinions of
other groups.

Per cent

100

All Judges
Librarians
Publishers
Others

r Nature of subject Unity of

Theme matter content

FIG. 2. An evaluation of the general aspects of content
FACTORS OF STYLE CONTRIBUTING TO READABILITY

According to the data presented in Table I, style ranks second
in importance among the four major categories when the opin-
ions of all judges are taken together. It may be recalled, how-
ever, that librarians as a group consider style the greatest con-
tributor to readability. This opinion is held by many individu-
als in all groups. For example, one publisher dismisses the en-
tire problem of readability in the following words: "What is a
readable book? It is a good story, well told." A readers' adviser
states, "When the reader of limited ability wants a readable
book, he wants a simple, pleasing style, an easy flow of expres-
sion/' Another says, "I feel very strongly that among the
qualities that contribute to readability, elusive qualities of
style .... and expression are foremost." This latter point of
view is expressed again by a director of adult classes as follows:
"For good reading, style is paramount/'

Most of these and similar statements are followed by a list of

WHAT IS A READABLE BOOK?

specific factors pertaining to style of expression and presenta-
tion which the writer considers important. As a result of the
emphasis which it receives, this category, designated No. Ill in
the classified list, contains the largest number of individual
factors.

When the total list was submitted to the judges for checking,
it was encouraging to discover a surprising agreement among all

TABLE IV

FACTORS RELATED TO STYLE OF EXPRESSION AND PRESENTATION RANKED OF

GREATEST AND LEAST IMPORTANCE TO READABILITY BY

EACH GROUP OF JUDGES

Factors Ranked in Highest One-
Fourth of 116 Factors

Factors Ranked in Lowest One-
Fourth of 116 Factors

it Adult vocabulary

yd Lucid, clear presentation

af Sentences not too involved

80? Start with the familiar

50 Enthusiastic attitude of the author

8 Adult approach

ik Non-technical vocabulary

4<r Chapters stimulating at beginning

a Sentences varied in length

ja Direct presentation

is Informal vocabulary

4^ Chapters promising at end

6a Narrative style

6c Descriptive style

3^ Paragraphs varied in length

4^ Clearcut chapters

ih Short words

ir Non-classical vocabulary

jo Distinguished style

5/ Emotional, sentimental attitude of
author

in Vernacular (even colloquial) vocabu-
lary

6e Poetic style

7^ Charming style

y/ Picturesque style

87 Parables

ib Vocabulary limited to 1000-1500
words

le Vocabulary easy enough for 12-14-
year-old child

$c Moralizing attitude of author

\d Exaggeration

8z Phantasy

groups regarding the factors of style which they believe are im-
portant in readable material. Table IV shows the factors of
greatest and least importance according to the combined opin-
ion of the three groups of judges. In Appendix B, Tables LXIX,
LXX, and LXXI, is given the opinion of each individual group
concerning important factors of style in a readable book.

The weight of opinion of all judges is that an informal (u),
non- technical (i&) 3 adult (if) vocabulary is an important con-

38 WHAT MAKES A BOOK READABLE

tribution to readability, whereas a vocabulary limited to 1,000-
1,500 words (i), easy enough for a 12-1 4-year-old child (i<?), is
not essential. Neither is a vocabulary of short (lA), non-classi-
cal words (ir), expressed in the vernacular (in) essential. These
opinions are in harmony with the consensus of workers with
adults. This group holds that the 4,000 most commonly used
words found in standard word lists should certainly furnish the
foundation of materials written for adults of limited reading
ability. They believe it is inadvisable, on the other hand, to
confine the vocabularies to these lists at the expense of adult
words within the experience of the reader. 10 Contrary opinion is
held by the advocates of an experimental vocabulary, such as
Basic English^ which is designed "to make it possible to say
almost everything we normally desire to say in 850 words." 11

With respect to the importance of factors which are classified
as stylistic devices, judges agree very generally that readable
material should have an adult approach (8#), starting with what
is familiar (8w) and within the reader's scale (8^). Other de-
vices of style, such as exaggeration (8^/), parables (87), and
phantasy (82), were either left unchecked or were checked and
qualified by such statements as "if appropriate to the subject,"
"if the subject-matter requires," or "depending on what the
author is attempting to do." Judges might well have added "if
we know what kind of reader finds these devices an aid to read-
ability." For the basic consideration in determining readability
of a particular book is whether a particular reader finds that
book easy and pleasant to read.

THE RELATIVE IMPORTANCE OF GENERAL ASPECTS OF STYLE

By the average rating of all groups of judges shown in Table
V and Figure 3, vocabulary, No. i,is given the highest value
among the eight aspects of style, and chapters, No. 4, the lowest
value. A marked similarity is observable among the mean val-

* Chancellor, op. cit., p. 16,

11 C K. Ogdcn, Basic English, London: Kegati Paul, Trench, Trubner & Co,, 1932,
Pp. 96.

WHAT IS A READABLE BOOK?

ues assigned by the three groups. Vocabulary ranks first in all
cases. Style of presentation generally takes precedence over
stylistic devices, which take precedence over sentences. They
in turn tend to rank above attitude of the author. This aspect

TABLE V

SUMMARY OF JUDGMENT CONCERNING THE RELATIVE INFLUENCE ON READABILITY
OF GENERAL ASPECTS OF STYLE OF EXPRESSION AND PRESENTATION

GENERAL ASPECT

ALL PERSONS

LIBRARIANS

PUBLISHERS

OTHERS INTER-
ESTED IN ADULT
EDUCATION

Range

20.43
11.38
8.29
6.63
12.32
10.43
17. oo
13.51

8. to
4-3*
4-05
3-70
7-09
5.61
6,25
6.40

50-8
25-5
25-0

20-0
30-0
20-0
40-S
30-0

20.00
9.96

r-ss
6.16
13.81
11.25
17.42
13.81

8.38
3.04
3. ii
3-90
7.53
5.76
5.98
6.50

SO-8
20-5
IS-o

20-0
3O-0
20-0
30-IO

30-3

*3.50
w-7S
8.33
5-75
10.17
9-S8
17.00
12.92

8.98
3-70
3-23
2.97
7-57
5-49
8.31
7.08

40-12
20-5
15-3

IO-O

25-0

20-0

40-5
30-0

18.50
12.25
9.20
7.64
11-95
9.91
I6.&
13.86

5-34
5-45
5.12
3.56
5-95
5.6i
5.62
5-53

30-10
45-5
2S-5
15-0
25-3

2O-0

26-5
25-5

3. Paragraphs

5. Attitude of author

6. Method of presentation. .
7. Style of presentation ....

ICO

Per cent
All Judges
Librarians
Publishers
Others

> I I 1 3 a

FIG. 3. An evaluation of general aspects of style

of style is generally conceded more important than method of
presentation. Since vocabulary and sentences are rated by most
judges above other aspects pertaining to mode of expression, it
appears reasonable to suppose that they are considered the basic
elements of expression, which, if readable, insure readable para-
graphs and chapters.

WHAT MAKES A BOOK READABLE

That the members of all groups agree fairly well in their eval-
uation of the various aspects of style is indicated by the stand-
ard deviations j which are generally smaller than those found for
aspects of content. There is evidence in the range of values,
however 3 to show that in every group one or more judges give
zero values to certain aspects, to which others give values of 20
or 30 per cent. It is such discrepancies as these that point to the
need of detailed study in order to determine for whom a par-
ticular style is readable.

FACTORS OF FORMAT IN A READABLE BOOK

Table VI lists the factors of format checked with greatest and
least frequency by all groups of judges. Factors held of greatest

TABLE VI

FACTORS RELATED TO FORMAT RANKED OF GREATEST AND LEAST
IMPORTANCE TO READABILITY BY EACH GROUP OF JUDGES

Factors Ranked in Highest One-
Fourth of 90 Factors

%b Attractive binding

9; Appropriate illustrations

4k Well-spaced type

9/ Maps and diagrams

4C Clear, legible type

7*7 Attractive page

ib Book of average size

i/ Light-weight book

3^ Dull-surfaced paper

4tn Black ink

9 Captioned illustrations

4/> Attractive type

8<z Sturdy binding

Factors Ranked in Lowest One-
Fourth of 90 Factors

ic Size of book larger than

a textbook
4?z Dull ink
40 Large type
4^ Small type

and least importance by separate groups are shown in Tables
LXXII, LXXIII, and LXXIV, Appendix B. Obviously, the
majority of judges believe that readers of limited ability will
find a book readable if it is of average size (ib] and light in
weight (if). It is, as Robinson has said, "a good little book,
easy to slip into one's pocket." Its binding is sturdy (80) and

WHAT IS A READABLE BOOK? 4 i

attractive (8). Its captioned illustrations (9^), maps and dia-
grams (9/) are appropriate for the content (97). It is printed on
dull-surfaced paper (3^) in a type that is well-spaced (4^), clear,
legible (40), and attractive (4^).

But these factors are not particularized. If, then, we dis-
cover that they actually do make a book readable for readers of
limited ability, we shall still need to determine what kind of
binding is attractive, what size of type is most legible, what are
the characteristics of an attractive page, and how large is a com-
fortable book. On such issues as these we find a diversity of
opinion. For example, an attractive page may depend in part
on color of paper. But what is the best color of paper? The
majority of librarians believe that non- white paper promotes
readability; whereas publishers and others interested in adult
education favor white paper. The latter opinion probably comes
closer to the truth of the matter, inasmuch as there is some
evidence to show that white, grayish, yellow, and red is the best
order of color for legibility. 12

Again, the three groups of judges agree that type in a read-
able book should be well-spaced (4^), attractive (4/>), clear and
legible (40). How can these qualities be attained? Individual
judges favor large type, small type, Granjon type, double-
spaced type, ii-point type, 12-14-point type, and type with
4-point leading. Generally, however, they agree (Table VI) that
neither large (40) nor small (4^) type is of any special impor-
tance to readability, provided that the product obtained is at-
tractive, clear, and legible. Disposal of the question in this fash-
ion shows the apparent need of discovering what is the range
of the optimum size of type which readers of limited ability
find readable. At the present time scientific evidence bearing
on the problem of best size of type is too conflicting and incon-
clusive to warrant even tentative conclusions. 13

H. Griffing and S. I. Franz, quoted by Madeline D. Vernon, The Experimental
Study of 'Reading (Cambridge: University Press, 1931), p. 170.

</., pp. 165-66.

4 2

WHAT MAKES A BOOK READABLE

THE RELATIVE IMPORTANCE OF GENERAL
ASPECTS OF FORMAT

Table VII summarizes the evaluation of nine aspects of for-
mat. In general the data are in harmony with the facts already
presented. Kind of type and printing (No. 4) is given the high-
est mean value for all groups. Factors belonging to this class
are agreed upon more frequently than those of any other
class, as shown in the first column of Tables LXXII to
LXXIV. Furthermore, a larger number of them are checked

TABLE VII

SUMMARY- or JUDGMENT CONCERNING THE RELATIVE INFLUENCE ON
READABILITY OF GENERAL ASPECTS OF FORMAT

GENERAL ASPECT

ALL PERSONS

LIBRARIANS

PUBLISHERS

OTHERS INTER-
ESTED IN ADULT
EDUCATION

Range

12.57
7,08
9-31
23.21

13-07
7.90
13-97

7-55
S-49
4.27
10.35
4.46
3-77

8.60

50-5
25-0

20-O

60-5

2O-O

25-0
45-0

20-0

45-0

13-32
7-35
9.29
23.16
4.58
7-13

14-94
8.13

12.10

6.10

6.20

4-53
H-93
3-77
4-54

8.84
i-48
6.78

30-5
25-0

20-0
609

15-0

25-0

40-0

20-O

30-0

8.66
4-50
9-92
26.04

s 9 .if

12.25
7-75
13.04

4-U
4-31
2.91

10.53
5-13
11.44

5:3

4.82

20-5

IO-0

17-5
40-15
20-4
10-4

30-4
15-5

2O-IO

14.31
8.05
9.18
21.73
S.4I
o.oo

n.73
7-59
16.00

9.89
4.82

4.66
7.30
3.7
3.01

8.80
3-31
7-40

50-5
18-0

20-5

40-5
iS-o

I2-O

45-o
15-0

2S-S

2 Number of pages .

3 Quality of paper.

4. Kind of type and printing

7. General appearance of

8. Binding

by occasional judges., as shown in the second column of the same
tables. It seems probable that the judges who have definite
views about size of type are responsible for the wide diversity
of opinion within the various groups that is indicated in Table
VIL This diversity may be noted in the wide range of value and
the comparatively large standard deviation of the mean for all
groups. Since both of these measures are largest among librari-
ans, it is clear that this group is not in close agreement as to
what is the relative influence of type on readability.

Librarians and others interested in adult education rank
length of line of least influence with notable consistency (Fig. 4).
Publishers, however, judge number of pages of least significance,
perhaps because they lack direct contact with readers of limited

WHAT IS A READABLE BOOK?

ability and are therefore not aware that length of line influences
book selection.

An examination of range of values assigned by various groups
shows that all publishers are more inclined to give some credit
to each general aspect than are members of any other group.
Although some publishers evaluate particular aspects as low as
4 per cent, certain judges in other groups consider them of zero
influence. Librarians especially tend to dismiss them as being
of no consequence, with the result that one or more of these

Per cent
All Judges
Librarians
Publishers
Others

100

1 lift if rll 111 I

to O* W H-5SO pqs

FIG. 4. An evaluation of general aspects of format

judges give zero value to all but two of the nine general aspects.
We may expect publishers to remain more open-minded toward
all aspects of format than other groups, since it is the category
most directly under their control.

WHAT FACTORS OF ORGANIZATION INFLUENCE READABILITY?

Data presented in Table VIII show the common opinion of
librarians, publishers, and others concerning the factors of or-
ganization that are of greatest and least importance to read-
ability. These data are derived from Tables LXXV, LXXVI,
and LXXVII, in Appendix B.

Only thirty-seven factors were suggested originally as prob-
able contributors to readability. Hence the number ranked in
the highest and lowest one-fourth by all groups of judges is
necessarily small.

44 WHAT MAKES A BOOK READABLE

A readable book, according to the judges, has three foremost
characteristics of organization: a striking title (ia)> a table of
contents (4*2), and descriptive chapter headings (ia). In addi-
tion, it has two other characteristics of almost equal importance:
paragraph divisions "not like a textbook" (^/), and an index
"with a catchy title" (4^). Obviously, what the judges mean is
this: "Use a little strategy. Put a book together so it looks
readable. Advertise to the prospective reader what he will find
in the book. And you challenge him to read it." Although other

TABLE VIII

FACTORS RELATED TO GENERAL FEATURES OF ORGANIZATION

RANKED OP GREATEST AND LEAST IMPORTANCE TO

READABILITY BY EACH GROUP OF JUDGES

Factors Ranked in Highest One-
Fourth of 37 Factors

ia Descriptive chapter headings

4# Table of contents

la Striking tide of book

3f Paragraph divisions not like

a textbook
4^ Index with catchy title

Factors Ranked in Lowest One-
Fourth of 37 Factors

4/ Index

judges than librarians and publishers favor interesting subheads
in bold-faced type, these two groups consider the factor of little
importance. Apparently, they fear that the practice may tend
to produce a textbookish product which does not look readable
to the adult of limited reading ability.

What to do with references in a readable book seems to be
a question open to disagreement (Tables LXXV-LXXVII).
Opinion is more diversified with respect to this aspect of organi-
zation than to any other. The weight of opinion among librari-
ans is toward references following the text. Publishers are about
equally divided on placing references in the text proper or in
the appendix, while the third group of judges comes out quite
uniformly for placing them in the text proper. One judge quali-
fied his first recommendation by a second that there be "few

WHAT IS A READABLE BOOK?

references. 7 ' It is clear that evidence is needed to prove whether
the placement of references influences readability for particular
kinds of readers.

THE RELATIVE IMPORTANCE OF GENERAL
ASPECTS OF ORGANIZATION

Table IX and Figure 5 show that all judges hold reference
guides as an important aspect of organization. Librarians and
publishers assign to this item a slightly higher mean value than to

TABLE IX

SUMMARY OF JUDGMENT CONCERNING THE RELATIVE INFLUENCE ON READABILITY
OF GENERAL ASPECTS OF ORGANIZATION

GENERAL ASPECT

ALL PERSONS

LIBRARIANS

PUBLISHERS

OTHERS INTER-
ESTED IN ADULT
EDUCATION

Range

I. Title of book

28 25
20.82
21.80
29.12

16.31
8.85
11.31
16.24

80-0
$0-7
SO-o
75-0

27-50
18.37
20.47
33.67

13-77
7.03
12.08
18.47

50-0
30-10
50-0

75-10

25-33
21.58
24.08
29.00

13.60
7 .8l
9.63
IO.S3

50-0
45-H
40-10
42-5

27.68

24-45
23.00
24.86

17.74
10.36
11.03
14.10

75-5
50-8
50-10
65-5

2. Chapter divisions

3 Paragraph divisions

4. Reference guides

other aspects. Other persons interested in adult education rank
the title of a book of considerably greater importance than refer-
ence guides. Lack of agreement within separate groups is indicat-
ed by the wide range of value and the large standard deviations
of the different means. Disagreement is also marked in the case
of titles of books which librarians and publishers evaluate any-
where from zero to 50 per cent, and other persons, from 5 to 75
per cent. Probably, it is more significant that the majority of
all judges agree upon a striking title as a mark of a readable
book than that individual evaluation of this aspect varies from
zero to 75 per cent.

OBTAINING REACTIONS FROM LIBRARY READERS

The next step in this inquiry has aimed to obtain reactions
from readers in libraries concerning what makes a particular
book readable. About forty readers' advisers in as many large

4 6

WHAT MAKES A BOOK READABLE

libraries of the country co-operated. Each was requested to ob-
tain through personal conferences with adult patrons, preferably
those of average or low reading achievement, as complete a
statement as possible of the factors they, believe make for read-
ability of a book recently read. Two other types of information
were desired: personal data concerning the reader, such as age,
sex, occupation, last grade in school, kind of material usually
read, and time devoted to reading per day; and the adviser's
own statement of the book's readable qualities.

100

Chapter
divisions

Paragraph
divisions

Reference
guides

Title of book

FIG. 5. An evaluation of general aspects of organization

For obtaining the reader's reaction to a book, the following
questions were suggested:

I. Why did you read this book?
a. How well did you like it?

3. Why did you like (or not like) the book?

4. Was the book readable; that is, was it easy and pleasant to read, and did it
bring interest and satisfaction?

5. What was there about the book which made it readable for you; that is,
that made it easy and pleasant to read?

Statements received from the readers in answer to these
questions were interpreted so far as possible in terms of the
classified list used previously. Comparisons were then drawn
between the reader's reaction to a particular book and that of
the readers' adviser to the same book. This was done for the
purpose of determining similarities and differences of opinion
relative to factors of readability found. Illustrations of these
similarities and differences will be given later.

WHAT IS A READABLE BOOK? 47

THE KIND OF READERS REPORTING

Reports from 170 readers' advisers represented the opinion
of as many readers, 74 men and 96 women. Their ages ranged
from fifteen years for one reader to eighty years for another.
The mean age was 31.4 years. Thirty- four of the readers were
unemployed, the remainder reporting a total of 56 different oc-
cupations. Included in this number were 28 home-makers, 12
stenographers, 8 students, 5 housemaids, and 5 secretaries. The
rest were scattered singly or in groups of two to four in other
occupations. Educational background, as measured by the last
grade attended in school, ranged from third grade to post-gradu-
ate in college. One person reported each extreme. Twenty-eight
of the readers claimed some education beyond high school. The
mean last year attended by the 156 readers who gave this infor-
mation was 10.6, representing an educational background some-
what beyond what was most desired for the inquiry.

The type of reading usually done by the 170 readers was clas-
sified into 44 categories, several individuals reporting that they
were accustomed to read a variety of content. It was evident
from the classified list that the reading tastes and habits of these
readers were widely diversified. Fifty-two persons reported they
read fiction most commonly; 29 reported popular non-fiction;
26, biography; 24, all kinds of reading; and 18, travel. The fol-
lowing kinds of material were reported as the usual choice of one
reader each: science, arts, engineering, short stories, useful arts,
mechanics, memoirs, adult education, inspirational books, and
radical books on religion and capitalism.

The amount of time devoted to reading per day was fre-
quently indicated by such terms as "little," "constantly/' "ir-
regular," "over an hour," and "evenings." Twenty-seven read-
ers claimed one hour of reading per day and a similar number,
two hours. Eleven averaged between two and three hours daily,
and 8, three hours. Various other amounts were reported up to
eight hours, which was recorded by one reader. The wide di-
versity in this information may have been due in part to differ-
ent reading habits, but probably much of it was due to diffi-

48 WHAT MAKES A BOOK READABLE

culty in estimating the amount of time usually devoted to
reading.

It is obvious from these data that we have obtained no syn-
thetic portrait of the reader of limited reading ability. We have
made no attempt, therefore, to present the opinions of this type
of reader. Neither have we generalized as to the factors which
different sorts of readers believe contribute to readability for
different sorts of reading, as we had hoped. The number of
readers proved too small to give a representative sampling of
any one class. What we have attempted is, first, to determine
the factors most frequently mentioned by all readers, regardless
of age, sex, educational background, and reading habits; and,
second, to compare them with the factors listed by the readers'
advisers for the same books. The extent to which the readers
and their advisers attributed readability of the books reported
to the same or different factors will be shown in succeeding para-
graphs.

AN ANALYSIS OF READERS* REACTIONS

The number of different books reported was 153. Seventeen
were read by more than one reader, The Good Earth having been
read by six persons. Some errors in interpretation doubtless oc-
curred for the reason that the particular factor to which refer-
ence was made was occasionally vague. For example, a colored
man, a freight-handler, who had left school in the fourth grade,
declared that Robinson Crusoe was the best book he had ever
read, because "the words were so you could read right along."
Whether he meant that the words are short, or easy, or familiar,
or non-technical, or informal, or colloquial could only be implied
from a knowledge of the book. Again, a housekeeper who had
attended school only to the sixth grade liked The Log Cabin Lady
so well that she read the book three times. She considered it
readable because "it follows along after you get started without
a break." This statement might be interpreted to mean that
the paragraphs are "progressively continuous," or that the style
of presentation is "rapid" or "easy and fluent," or that the ap-
peal is "within the reader's scale." When the meaning of a

WHAT IS A READABLE BOOK? 49

statement could not be determined with some definiteness from
a reader's total reaction to the book it was omitted from classi-
fication.

Other persons gave clear reasons for considering a particular
book readable. A college graduate, a field director of boy
scouts, who is especially interested in non-fiction, found The
Epic of America very readable. He attributed its readability to
the following factors: large easy type, good paragraphing, com-
fortable size, clear-cut presentation, vivid introduction, pro-
logue, and fair interpretation of people and events. Such factors
could be readily classified.

When the opinions of all 170 readers had been interpreted
and tabulated, it was found that factors pertaining to style had
been mentioned 193 times. Factors of content ranked second
with 151 mentions; format, third, with 78; and factors of or-
ganization, last, with a total of 8 mentions. A distribution of the
114 different factors mentioned by the readers showed that of
the 28 which were most frequently mentioned as contributing
to the readability of a particular book, 13 are aspects of style,
9 of content, and 6 of format.

Many of these factors which are listed in Table X were trans-
lated from such personal statements as: "I was interested in
finding out how to cure sleeplessness" (IV, 2/) 3 "I always want
to read about India" (IV, la), "I learned how to speak correct-
ly" (IV, 2p), and "I felt as though the author meant what he
wrote" (III, 5J-). Statements of this sort were noticeably pre-
dominant among readers of less than average schooling and of
somewhat mediocre reading tastes. They probably do not mean
that factors of format and organization exercise no influence on
readability, but, rather, that the influence, though probably
less than that of style and content, is so subtle that the reader is
conscious only of interesting content written in a manner he
can understand. There is need for investigation to determine
the extent to which attractive format creates interest in books
and adds to readability for different classes of readers.

Another frequent type of comment, and one which defied

WHAT MAKES A BOOK READABLE

precise classification, was that a particular book was considered
"easy/ 7 "easy to read/' "easy to understand," or "not very
hard." In some cases, the readers* adviser clarified such state-

TABLE X

FACTORS MENTIONED WITH GREATEST FREQUENCY BY READERS AND READERS'
ADVISERS IN COMMENTS REGARDING THE READABILITY OP THE SAME BOOKS

Factors Ranked in Highest One-Fourth
of 114 Factors by Readers

Factors Ranked in Highest One-Fourth
of 104 Factors by Readers* Advisers

IV, ij Interesting subject matter
IV, la Theme what people want to

read about
*IV, 10 Real or ideal life experience

III, *] a Simple style

IV, 2p Informational material
*III, jk Easy, fluent style

IV, ic Familiar content

*I, a* Brief book

*I, 40 Clear, legible type

*I, 44 Large type
*III, 7^ Vivid, colorful style

*I, ^q Good print
*III, 7,g Entertaining style
III, ic Easy vocabulary
III, ik Non-technical vocabulary

III, 5| Sincere attitude of author

IV, ii Theme people and personalities
*IV, i/ Theme romance and action

IV, 2/ Helpful subject matter
IV, 2r Satisfying subject matter
III, ij Common, familiar vocabulary
*III, yd Lucid, clear style
III, 7? Natural style
III, 8 Omission of nonessentials
*I, ig Book of comfortable size
*I, 9; Appropriate illustrations
III, 6b Descriptive style
III, 7* Direct style

*I, 4^ Good print

*I, 40 Clear, legible type

*I, ig Book of comfortable size

I, i/ Light-weight book
*l y i& Brief book
*I, 40 Large type

I, la Small book
II, 2<z Descriptive chapter headings

III, ig Simple words
*III, jk Easy, fluent style

IV, i Theme human interest
*IV, 10 Real or ideal life experience

I, 83 Attractive binding
I, 9/fc Attractive illustrations
*III, 7| Entertaining style
III, jp Informal style
I, i if Book of average size
I, 6a Wide, liberal margins
*III, 7^ Vivid, colorful style
III, jj Simple style
*IV, i/ Romance and action
I, 7^ Attractive page
*I, $j Appropriate illustrations
*III, jd Lucid, clear style
III, le Simple sentences
III, Sf Conversation

* Factor mentioned by both readers and readers* advisers.

ments by attributing the readability of the book to "short sen-
tences/' "lack of technical words/' "use of first person singu-
lar/' or "simple language" qualities that presumably relate
to ease of reading and ease of understanding. In other cases,

WHAT IS A READABLE BOOK? 51

the readers' adviser ascribed none of the readability of the book
to elements that may promote ease.

From the records of readers' advisers it was found that fac-
tors of format were mentioned 223 times; those of style, 162
times; of content, 65 times; and of organization, 46 times. Dif-
ferent factors numbered 104, their frequency of mention ranging
between 37 and i . For the sake of convenience in comparison,
the most frequently mentioned factors of readability obtained
from the advisers' reports are also listed in Table X. It may be
noted that of these 26 factors, 13 pertain to format, 9 to style, 3
to content, and i to organization, in the same order as the total
number of mentions. Factors mentioned both by readers and
readers' advisers are indicated in the table by asterisks.

That readers' advisers should most often attribute the reada-
bility or non-readability of 153 books to factors of format is
perplexing, when one recalls that in the earlier part of the in-
quiry librarians and advisers rank this category third in value
among the four major categories. Furthermore, data presented
in Table VII show that the ranking has been done with con-
siderable agreement. Several assumptions may be made. One
is that factors of format are actually most influential for reada-
bility or non-readability of the particular books read, despite
the fact that readers themselves tend to minimize the effect of
format.

A more plausible assumption is that in the earlier inquiry
readers' advisers were judging readability in general. Here they
are judging the readability of a specific book, which may or may
not be the same thing. It may be easier to talk about interesting
content, pleasing style, and other more or less subjective quali-
ties than it is to label them definitely. On the other hand, good
print, large type, light weight, brevity, and comfortable size are
relatively constant and objective for any one judge, and conse-
quently more freely mentioned in evaluating a particular book.

The third assumption, and one on which some evidence is
available, is that readers' advisers may not have read some of
the books. Hence, they felt unqualified to comment on factors

52 WHAT MAKES A BOOK READABLE

of content and style, confining their evaluation, instead, to fac-
tors of format, which can be observed with little trouble, yet
with considerable accuracy. Evidence to support this assump-
tion was found in the reports of six books which readers' ad-
visers said they knew only from handling, not from reading. Six
other books were given no evaluation, presumably because they
were quite unfamiliar to the readers' advisers. It seems prob-
able that all three of these assumptions may explain in part the
difference in emphasis noted between the general evaluation of
readability made by the readers' advisers and their specific eval-
uation of particular books.

The twelve factors most frequently mentioned by both read-
ers and readers' advisers may be noted in Table X. In ascribing
readability to format, both groups most frequently state that
the book is of comfortable size (I, ig), is brief (I, 2#), with large
(I, 4#) clear, legible (I, 40) type, good print (I, 4^), and appro-
priate illustrations (I, gf). The style of a readable book is gen-
erally vivid and colorful (III, 7^), lucid and clear (III, jd), easy
and fluent (III, 7&), and entertaining (III> Jg) ; and its content
is based on real or ideal life-experience (IV, 10), or on romance
and action (IV, i/).

In order to show concretely the similarities and differences
found between reactions of the reader and of the readers* ad-
viser to the same book, typical statements are cited in the out-
line on pages 53 and 54. They show the marked tendency of the
readers* adviser to point out desirable factors of format and of
the reader to emphasize style and content. All statements are
quoted exactly from the reports of the two groups.

WHAT IS A READABLE BOOK?

TYPICAL REACTIONS OF READERS AND READERS' ADVISERS
TO THE SAME BOOK

BOOK

Jackson, J. A.,
and Salisbury,
Helen, Outwitting
Our Nerves

Swift, Psychology
and the Day's
Work

Fernald, Expres-
sive English

READER

REACTION or READER

grade education.
Reads novels
usually.

Female 27 yrs. It had a whole chapter
Mother. Eighth- on being unable to
sleep, and other things
I needed. I could find
anything I wanted in the
book after I read it. So
many times I wanted
to show my husband
something and I could
turn to it in a second
because everything im-
portant in it was
brought out.

Male 46 yrs.
Unemployed.
Eighth grade.
Reads psychol-
ogy, religion, etc.

Male 35 yrs.
Carpenter.
Sixth grade.
Reads newspa-
pers and maga-

Yes, the book was
readable. I don't know
why.

It was easy to under-
stand. Well expressed,
Good print. Light to
hold (to read on the
cars).

REACTION OF
READERS' ADVISER

Medium-sized, light,
good print. Well or-
ganized. Each chapter
has several headings of
different kinds. Para-
graphs are emphasized
by black ink sugges-
tions. Most of the
chapters are summa-
rized. There is a glos-
sary, a bibliography,
and an index. The con-
tent satisfies.

Non-technical, brief
presentation of the re-
lation of psychology to
everyday life. Good
print. Handy size.
Journalistic style.

Information given In
an interesting manner.
Many apt examples.
Convenient size. Clear
print.

Hugo, Le s Misera- Male 20 yrs.

ble$ t Adapted by Unemployed.

Ettie Lee Fourth grade.

Reads newspa-
pers.

Gide, First Prin-
ciples of Political
Economy

It was exciting and Short sentences. Sim-
true. It was not too pie language. A great
long and the print was deal of action,
good.

Male 35 yrs.
Unemployed.
Sixth grade.
Reads some
newspapers and
the Pathfinder.

I couldn't understand
it. Too many big
words. I couldn't "fig-
ger out" what he was
talking about. He used
so many big words that
I never seen. I tried to
use the kid's dictionary
but by the time I'd
hunted up the words
I'd clean forgot what
he was talking about.

Admirable in format,
etc. Vocabulary too
difficult for the level of
the subject matter.
Here is a case that
might pay to translate
into "Brief English."
Subject matter is too
elementary for those
who can read it readily,
on account of vocabu-
lary.

WHAT MAKES A BOOK READABLE

TYPICAL REACTIONS OF READERS AND READERS' ADVISERS
TO THE SAME BOOK Continued

BOOK.

Humphrey, Story
of Mans Mind

READER

Female 40 yrs.
Social worker.
H.S. graduate.
Reads biogra-
phy and social
subjects.

REACTION OF READER

Simply written so that
I could understand it.
Made me want to go
further in the subject.
Good format. Liked
the diagrams. Style
was readable. It re-
ferred to things I al-
ready knew about, not
over my head.

REACTION OF
READERS' ADVISER

Form of book looks
easy. It is simply writ-
ten without technical
vocabulary. Has sim-
ple illustrations and al-
lusions to everyday
things.

Defoe, Ro&inson
Crusoe

Male. Colored Crusoe was like me, Written in first person
30 yrs. Freight- he made the best of singular. Short chap-
handler. Fourth everything. The words ~~

were so you could read

right along.

grade. Reads
everything that
will "learn him
something."

ters. Proceeds without
a break. Content of
primitive living, strug-
gle for conquest, rich
use of ingenuity. Lack
of technical words.
Short sentences. Good-
sized print. Clear il-
lustrations. Paragraph
divisions. Subheads.

Halliburton, Female 26 yrs. Made things seem real. Author's enthusiastic

Royal Road to Ro- Knitter in a ho- Easy to read. Sim- manner. Attractive

mance sierymill. ply written. Easy to size and general or-

Eighth grade. understand. ganization.

Reads travel

books.

SUMMARY OF INQUIRY RELATIVE TO READABILITY

The evident similarity of opinion among the judges who eval-
uated the classified factors proves that publishers, librarians,
and others are more united in their thinking about the factors
which may influence readability than perhaps they themselves
realize. It has been shown that groups interested in the prob-
lem can analyze readability as a general quality, and that
they can do so with more than a fair degree of consistency.
Obviously, mere agreement of opinion does not establish that
opinion as truth. However, since many opinions compiled in
this inquiry are based upon careful observations of reading in-

WHAT IS A READABLE BOOK? 55

terests, habits, and demands of adults, it is highly probable that
those agreed to by a majority of judges are symptomatic of the
truth as it pertains to factors which make a book readable for
readers of limited ability.

The lack of agreement among individual judges is quite as
notable as the presence of agreement among groups of judges.
Wide diversity of opinion regarding the relative influence of
some general aspects is shown by a range in value from 10 to 80
per cent. For specific factors, the variability in judgment ranges
in several cases from double checks, indicative of special signifi-
cance, to no check, implying zero significance. We can no more
conclude that one is right and the other wrong than we can con-
clude that general agreement among the judges establishes a
factor as important or unimportant for readability. In both
cases we need supporting evidence to determine how a partic-
ular factor contributes generally to the readability of a certain '
kind of material for a certain class of reader.

Upon the question of what factors actually influence reada-
bility and of what is their relative importance in determining
the readability of a book, this inquiry can do little more than
suggest an answer. Its findings rather offer a challenge to at-
tempt to discover whether factors generally believed to influ-
ence readability really do so. For example, does provision for
certain aspects of format, such as average size and light weight
of a book; black ink; dull-surfaced paper; clear, legible, attrac-
tive type; captioned illustrations; and maps and diagrams con-
tribute to readability, as the judges believe? If so, what size of
type is clear, legible, and attractive? Is size of less significance
than leading or interlineage? Do captioned illustrations pro-
mote readability? If so, what kind is more effective, the car-
toon-type as suggested by eleven judges or photographs as sug-
gested by one judge?

Innumerable other questions arise from an examination of
the data obtained from the inquiry questions that can be an-
swered only by experimentation with the reader for whom
readable materials are apparently needed. Subjective testi-

56 WHAT MAKES A BOOK READABLE

mony of readers themselves regarding what makes a particular
book readable or non-readable is not enough. Evidence of an
objective sort should be sought in order to reveal factors of
which the reader may be quite unconscious.

If the investigation is to produce practical outcomes, it must
be resolved into major aspects. These must then be analyzed
into a series of related problems for detailed study. Research
directed toward the discovery of facts pertaining to format and
organization should aim to give a list of desirable practices that
will guarantee books that look readable, especially for adults of
limited education. Conrad's investigation of reading material
for native adult illiterates and near-illiterates is a promising
beginning. The study of reading content as attacked by Waples
and Tyler furnishes data by means of which librarians, writers,
and publishers may determine what different groups of adults
want to read about. From these data, too, publishers and writ-
ers may be guided in their further efforts to prepare informa-
tional material that different classes of readers will read if the
opportunity allows.

The fourth aspect of readability, and one which both judges
and readers agree is of major importance, is concerned with the
style of expression and presentation of material by means of
which content is made "easy and pleasant" for different classes
to read. It is with this aspect that we are concerned in the re-
maining chapters of our report. We have anticipated that a
thoroughgoing investigation of problems related to ease or diffi-
culty should ultimately supply needed standards for making
vocabulary, sentences, and paragraphs easy to read. It should
also discover the methods of presentation that tend to insure
pleasant and satisfying reading. When each aspect of readabili-
ty has been defined precisely for different readers, it seems rea-
sonable to believe that all the facts may be synthesized and
utilized in the production of more books which look readable
from the point of view of format and organization, and which
are readable if their content and style make them easy and
pleasant to read.

CHAPTER III
HOW WELL DO ADULTS READ?

A WORDING to the plan outlined in chapter i, only a
single aspect of readability has been studied intensively
for this initial report of what makes a book readable.
That aspect is ease or difficulty of materials comprising the gen-
eral reading of adults. Our primary concern is to discover the
elements inherent in such materials which influence ease or
difficulty.

It must be recognized at the outset, however, that difficulty
is a relative quality and that it can be defined only in terms of a
reader's comprehension. Material is easy for one reader because
he can understand it, whereas it is difficult for another reader
because he cannot understand it. This phenomenon raises two
practical questions: For what class is it most important to iden-
tify elements of difficulty? And, How well are readers of that
class able to read the kind of materials in which elements of
difficulty need to be defined? Inasmuch as it is the reader of
limited ability who finds most of the material now written for
adults too difficult, it is for this class of reader that we have
aimed to identify elements which influence difficulty.

The purpose of this chapter is, first, to show how well repre-
sentative groups of adults are able to read, and, second, to de-
fine more precisely who are readers of limited ability. In ac-
complishing these purposes, it was recognized that certain ob-
jective measuring instruments are essential to an accurate de-
termination of reading ability. A representative sampling of the
adult population is also necessary in order that evidence con-
cerning how well adults read may be as valid as possible. A
description of the testing instruments used, together with a
characterization of the groups tested, is presented in the sec-
tions that follow.

58 WHAT MAKES A BOOK READABLE

Since testing conditions vary widely in different situations,
it was impossible to measure the ability of all groups by all the
testing devices used in the study. For example, some groups
were tested for ability to comprehend the content of general
magazines, books, and newspapers commonly read by adults.
Others were tested for speed and accuracy of oral reading. All
groups were tested for ability in silent reading as expressed in
terms of grade norms. In view of these variations, the evidence
concerning reading ability for all groups could not be considered
collectively. Instead, it has been compiled and presented in a
manner that provides the most satisfactory basis for interpret-
ing facts obtained from single tests. How well different classes
of adults read general material is shown first. Their ability in si-
lent reading is revealed next in terms of grade norms. Their
speed and accuracy of oral reading is presented last.

HOW TO MEASURE COMPREHENSION OF GENERAL
READING MATERIAL

An examination of existing tests in reading showed that none
was available for measuring comprehension of general reading
materials such as are found in books, magazines, and newspa-
pers. For this reason the preliminary task of constructing such
a test was undertaken. The type constructed was determined in
the light of the theory generally accepted regarding the nature
of comprehension. That comprehension is not a single unitary
process is commonly conceded. It is, rather, a blending of many
processes whose totality represents understanding. But com-
prehension has not been resolved satisfactorily into its compo-
nent processes. Consequently, first one and then another lias
been isolated and taken as a measure of comprehension.

In selecting aspects of comprehension to use in constructing
reading tests, we have made two assumptions. The first is that
ability to grasp the essential meaning of a selection in the form
of a general impression is the outcome most frequently demand-
ed of adults in reading general material. The second is that a
large amount of reading is done in order to gain specific infer-

HOW WELL DO ADULTS READ? 59

mation contained in the selection. Furthermore, since ability
to grasp the essential meaning seems to depend in a degree upon
ability to react satisfactorily to specific elements contained in
the selection, it is assumed that any technique devised to meas-
ure the first outcome will indirectly measure the second.

THE ADULT READING TEST

The construction of the adult reading test used in measuring
comprehension of general materials is described in detail in
Appendix C. The test consists of two forms printed in separate
booklets. Form I contains a series of twenty-four unmutilated
paragraphs of fiction selected from magazines, books, and news-
papers. Form i contains a similar number of paragraphs of non-
fiction selected from the same general source. An attempt was
made in all cases to choose complete units of thought, the sense
of which could be given in single summary statements. In
order to increase the reliability of the measure of comprehen-
sion, two reactions were required. The first was the identification
of the best summary in a series of five statements relating to the
paragraph. The second was the recognition of a detail not found
in the paragraph. Reactions were indicated in the manner pre-
scribed by the following passage, quoted from the set of direc-
tions accompanying the tests:

You are to read each paragraph carefully. Then read the five sentences
below the paragraph. Put a CHECK MARK (V) before the sentence that tells
best what the paragraph said. One of the five sentences tells something that
is wrong or that is not in the paragraph. Mark that sentence with a ZERO (o).
Do not mark the other three sentences.

The nature of the test is apparent from the following sample
paragraph taken from the non-fiction test, designated as
Form 2.

SAMPLE PARAGRAPH

For nine years I have watched, during the spring and summer months, a
nighthawk that lives in the daytime on a big limb near the end of my garden.
He can usually be found at the same spot from April to September. It may be
that a member of a second or third generation now lives there; but it is truly
a family tree. I often wonder why a bird becomes attached to one place,

60 WHAT MAKES A BOOK READABLE

unless it is because it is contented there. Explain it as you will, contentment
results from peace of heart, and the home-loving heart is usually at peace.

PARAGRAPH SUMMARY

i. *' . . Ever since the nighthawk built his nest in my garden he has lived
there all the year round.

2. The tree where the nighthawk lives is truly a family tree.

3. The nighthawk can usually be found on the same limb during spring

and summer.

4. The nighthawk must be contented in my garden for he has lived

there from April to September for nine years.

5. l! The same nighthawk or perhaps a member of his family has lived

in my garden during spring and summer for nine years.

The highest possible score on the fiction test, Form i, is 92.2;
and on the non-fiction, Form 2, 91.4. Test scores attained on the
two forms by an individual or by a group are therefore com-
parable. 1

From the records secured by the use of these test-forms, sig-
nificant facts were obtained concerning the relative ease or diffi-
culty of fiction and non-fiction for general adult readers. Some
evidence was secured also with respect to the type of material
that adults read and interpret most readily.

SECURING GROUPS FOR TESTING

In presuming to answer the question of how well adults read,
we had no alternative but to test a limited number of individuals
and then to generalize on the results, applying them to the
larger population of adults in general. Although it is theo-
retically desirable to select a sampling in such a way that the
cases will represent fairly all portions of the population, it
is practically impossible to do so.

The obstacles to be met in a program designed to measure
adult abilities are many. In the first place, suitable testing op-
portunities are not readily available. At the time this study was
carried on, the most favorable conditions were found in evening

* The method of scoring the test, together with an evaluation of the test for relia-
bility and validity, is given in Appendix C.

HOW WELL DO ADULTS READ? 61

schools where students possessed a variety of abilities and stages
of educational progress. Only occasional groups of adults at-
tended day school before the organization of classes for emer-
gency relief. Under ordinary circumstances, such groups are
smaller in number and less regular in attendance than are those
in evening schools. Allowance must be made, therefore, for con-
siderable loss in time and materials if the testing program ex-
tends beyond one day. For example, a day-school group in Chi-
cago was tested on two different days, with the result that al-
though more than 40 adults took Form I of the Adult Reading
Test and a similar number took Form 2 three days later, only
12 persons were common to both groups and consequently
tested by both forms.

Other organized groups may be found in evening classes con-
ducted by religious organizations or in classes provided for in-
mates of penal institutions. In the latter case, especially, stu-
dents tend to be notably retarded and to possess obvious read-
ing deficiencies. This is due largely to the fact that for the most
part prison schools have attempted to do little more than to
correct glaring educational deficiencies.

Groups classified at the secondary-school and college levels
are usually available, and in the present study they proved of
great value. The inclusion of students at the former level
seems warranted, inasmuch as a large percentage of our popu^
lation who enter high school do not continue their education
farther. It seems reasonable to assume, therefore, that the read^
ing achievement of the present generation of high-school pupils
may indicate roughly how well some adults will read a decade,
or two hence. Unless they engage in adult educational activities'
in later years, it is probable that their present reading achieve-
ment is higher than it will be in the future.

One may question the inclusion of college groups in a testing"
program which is designed to discover readers of limited ability.
During the past several years, however, findings have been
obtained which show conclusively that college students not only
differ widely in reading achievement but that many of them ex-

62 WHAT MAKES A BOOK READABLE

hibit surprisingly immature and deficient reading habits. It
seems pretty clear that too much cannot be taken for granted
concerning the ability of college students to read either their
required course assignments or general adult books and periodi-
cals. By restricting the college groups tested to those whose
fading opportunities had been relatively meager, it seemed
probable that some would manifest limited ability.

Since parent-teacher organizations are usually of a hetero-
geneous character, they furnish desirable subjects for testing a
range of abilities, provided that they can be convinced the
project is worth while, or if a trade in benefits can somehow be
arranged. In the present study both conditions were met satis-
factorily.

A second obstacle to testing adults lies in the hesitance and
apprehension with which they approach a test of their abilities.
This reaction is especially characteristic of adults whose edu-
cation has been neglected or has been obtained in a language
other than the one used in the test. It is obviously essential
that harmonious relations be established between the adminis-
trator and test subjects if results are to be reliable. Various
methods were used to secure confidence and consistent effort.
For example, members of parent-teacher groups were fearful of
exposing reading deficiencies. They were allowed to draw num-
bers by which to designate their tests. Retarded adults were
constantly encouraged and were frequently reassured that ina-
bility to answer a test item or to finish a test was attended by
no serious consequence. This was done to promote confidence
and to prevent guessing, which would tend to invalidate the
findings. A knowledge of the real purpose of the tests probably
obviated the desire to finish at any cost and promoted honest
effort*

WHAT GROUPS WERE TESTED?

Table XI presents the name, geographical location, approxi-
mate grade-placement or last grade attended in school, and size
of the various groups tested by the adult test. They are ar-
ranged into six somewhat homogeneous classes with respect to

HOW WELL DO ADULTS READ?

educational advancement. The first two represented adults who
were continuing their education at the elementary level. The
South Carolina group included workers in textile mills whose
early schooling had been neglected. The next three groups,
made up of working men and women of limited education, rep-

TABLE XI

NAME, GRADE-PLACEMENT, GEOGRAPHICAL LOCATION, AND SIZE OF GROUPS TESTED

Name

Grade-
Placement

Geographical
Location

Number
of Cases

Elementary classes
Glenn Street School

Anderson S.C

Evening-school classes
Englewood

7, 8

Chicago, 111.

J Sterling Morton

7 8

Cicero 111.

Dante . .

Chicago, 111.

Junior high schools
Berea Foundation

7, 8, g

Berea, Ky.

A & M Model School

780

Tallahassee, Fla.

4.8

High schools
Berea Academy

Q, ICL II % 12

Berea, Ky.

Berry High School Freshmen .

Rome, Ga.

Rabun Gap-Nacoochee

IO, II

Rabun Gap, Ga.

4.O

A & M Model School

ID, II, 12

Tallahassee, Fla.

Berry High School, (Jr. and Sr.). . .
Colleges
Berea

ii, ia
Fr.,S.Jr.,Sr.

Rome, Ga.
Berea, Ky.

14
98

Berry

Fr., Jr.

Rome, Ga.

Non-school group
(parent-teacher associations)
Hinsdale . .

Hinsdale, 111.

Congress Park

Congress Park, 111.

Total

756

resented the elementary classes found in the metropolitan eve-
ning school.

The two junior high-school groups and the five high-school
groups listed in the table included young people and adults
whose secondary education had been delayed. For example, Be-
rea Foundation Junior High School aims to receive only "pupils
from the mountain counties of Appalachian America who are

64 WHAT MAKES A BOOK READABLE

sixteen years of age and over, who do not live within reasonable
reach of adequate public schools, and who have not completed
the first nine grades/' The junior and senior high-school groups
of the Model School of the Agricultural and Mechanical Arts
College, Tallahassee, Florida, enrols young people and under-
privileged adults of the colored race, some of whom are prepar-
ing for college. Berea Academy accepts students from a variety
of contributing schools. At the time this study was begun, the
entrance requirements were three units in the ninth grade. Some
exceptions were made, however. Martha Berry High School
offers educational opportunities to young people in the moun-
tain districts of northern Georgia. Rabun Gap-Nacoochee
School calls itself "a school that deals with mountain conditions
as an economic problem." Whole families from the mountain
regions of Georgia, North Carolina, and South Carolina attend
school there under a revolving farm plan, and pay for their
cottages and tuition with their produce. Other students help
the farmers two days a week during the school year and ten
weeks during the summer to pay for tuition and board.

The students of Berea College and Martha Berry College are
largely mountain people. Entrance requirements to these col-
leges are practically the same as those of the Southern Associa-
tion of Colleges.

The parent-teacher groups tested represented the better-edu-
cated American adults not in school. Although the members of
these organizations possessed a diversified educational back-
ground, most of them had attended high school. Many were
high-school graduates, some of whom had received more or less
college training.

Additional data regarding the personnel of the groups tested
were recorded on the test blanks by the individuals them-
selves. The validity of the data may therefore be open to ques-
tion. The main facts obtained are summarized as follows:

1. Of the 756 subjects whose reading ability was measured by
the adult reading tests, 416 were women and 340 were men.'

2. The age of the subjects ranged from 15 years to more than

HOW WELL DO ADULTS READ? 65

50 years, as shown in Table XII. The mean age was 22.7. The
size of the standard deviation, 7.5, indicates that for approxi-
mately the middle two- thirds of the subjects the age ranged be-
tween 30.2 years (22.7+7.5) an< i J 5- 2 years (22.77.5).

3. Educational training of the group, as measured by the last
grade attended in day school, ranged from third grade, reported
by one person, to above twelfth grade, reported by 246 persons

TABLE XII

NEAREST AGE IN YEARS OP 756 ADULT SUBJECTS

Nearest Age in Years

Number

Per Cent

124

16.4.0

'JQI

<I .72

1C. 21

<.O^

o w
or

1.7O

4.0

4.O

C..2Q

1.08

O.C.T

ro+

O.IT

Total

7C6

IOO.OO

22. 7 .18 Mean

(Table XIII). The mean was calculated at grade 11.01 by in-
terpreting "beyond 12" as 14. If the data regarding previous
schooling can be taken as reliable, it is evident that approxi-
mately the middle two-thirds of the adults tested on the adult
test ended their day schooling between grades 8.32 and 13.70.

4. Of the 756 subjects who gave information concerning at-
tendance at evening school, 660 reported from no attendance to
three months, whereas three claimed a period between five and
six years. The mean attendance was 3.1 months.

5. Amount of time devoted to reading per day was admitted
by the subjects as representing no more than a crude estimate.
Two persons reported that they never engaged in any reading.
At the opposite extreme of the scale were 73 persons whose

WHAT MAKES A BOOK READABLE

average reading time exceeded three hours per day. The mean
for the group was 1.56 hours.

6. The occupational classification used in Table XIV was
adapted from the report of the Bureau of the Census, and hence
tends to obscure the diversity of occupation actually obtained
from the records. An additional classification, "Educational/'

TABLE XIII

LAST GRADE ATTENDED IN DAY SCHOOL, AS REPORTED
BY 756 ADULT SUBJECTS

Last Grade Attended
in Day School

Number

Per Cent

o .

o n

O.7Q

i .72

i. 08

4.2

c.c6

II .<!

II QO

8.20

6.08

l<2

1.4-4.

IQ QC

Beyond 12*

246

?2.<4

Not reported .... *

O. <?

Total

756

IOO OO

1 1. oi .07 = Mean

* Beyond a wai interpreted as 14 in computing the mean and stand**
ard deviation.

was added to include persons attending school. Although this
group may seem disproportionately large, it is explained by the
size of the under-privileged groups attending opportunity
schools. Many adults were devoting a part of their time to a
trade, yet preferred to class themselves students.

As indicated by these descriptions, the various groups tested
included important elements in the adult reading public, if not
a representative sampling of it. It is clear that we have sampled
a wide range of educational levels from elementary to college,

HOW WELL DO ADULTS READ?

But it is also clear that these groups tend to be selective rather
than typical, that the individual members have educational in-
terests of some sort or they would not be found in classrooms or
in parent-teacher organizations,

No attempt was made to discover facts concerning illiterates.
Their needs are urgent, but they are not reading nor are they

TABLE XIV

OCCUPATIONS OF 756 ADULT SUBJECTS*

Occupation

Number

Per Cent

Agriculture

O C7

Building and construction

0.26

Educational study

<M<

72 OQ

Manufacturing and mechanical industries

c.i6

Transportation and communication

O.I?

Trade

2.78

Public service

O. <7

Professional service

3.84

Domestic and personal service

II. CI

Clerical occupations

1.72

Unskilled labor

O.4.O

Not reported

I.4<

Total

7C6

ICO. CO

* Classification was adapted from the report of the United States Bureau of Census.
See: The World Almanac and Book of Facts f of 1933. (New York: New York World Tele-
gram, 1933), P- 4S6.

generally simulating reading interests and habits. Only those
persons were tested who had learned to read and were apparent-
ly doing some reading. Although the findings obtained from
such a testing program are not conclusive, they probably are
suggestive of the level of reading ability possessed by the gen-
eral public who reads adult magazines, books, and newspapers.
Groups of adults tested by other types of tests will be de-
scribed in later sections.

HOW WELL DO ADULTS READ GENERAL
READING MATERIALS?

Table XV presents the distribution of scores on Forms I and 2
of the Adult Reading Test made by 756 adults. It gives also the

WHAT MAKES A BOOK READABLE

mean score and the standard deviation of the mean for each
test. For Form i, the fiction test, the mean score was 53.86; for
Form i y the non-fiction test, it was 47.86. The difference of

TABLE XV

NUMBER. AND PERCENTAGE OP 756 ADULTS MAKING VARIOUS SCORES ON THE
TESTS OF FICTION AND NON-FICTION

SCORES

FICTION TEST

NON-FICTION TEST

No.

Per Cent

No.

Per Cent

SQ.QC Q4..QC

2
2 9
70

55
64

5*
46
41
39
41
32
3i
ai
27

22
19

0.26

3-84
9.26
11.38

7-54
7.28

8.47
6.88
6.08

5-42
5.16

5-4*
4-23
4.10
2.78

3-57
2.91
2.51
i. 06
i. 06

-53
0.26

4
21
40
61
77
59
59
66

41
56
48
5^
53
43

13
20
12

2
2

o-53
2.78
5-29
8.07
10.19
7.80
7.80
8.73
5'4*
7-41

6.35

6.88
7.01
5.69
2.78
1.72
a. 65

i-59
0.66
0,26
0.26

8/1 QC 80 QC

7O QC 84. QC

7/1 or 7Q.QC

60 QC 7ji QC

u y yj /^"yj

6x QC 6Q QC

CO QC 64.. QC

fx QC CQ.QC

T-y-io jT"7j

To of 4/t,QC, . ,

74.. Q< 'JQ.QC

2o Q< 74. QC

0.4..QC aq.QC

TO QC 2.A.QZ

IA.QC IQ,Q<

Q Q< I4..QC

4.Q< 0.0<

.OC- 4.QC

A.Q/f .OC

. Q C 4..OC

I4.,Q< <J.O<. . . ,

10 QC IA QC

2,4. QC IQ.QC

2Q. QC-" 24.. Q<

0.13

Mean53.86-59
S.D. s 24.oi .42

47.86.5i

2I.24.37

6.00 between the means is statistically significant, since it is 7.6
times its probable error, ,79, The deduction to be drawn from
this fact is that adult readers of the kind used in this study com-
prehend fiction better than non-fiction, when comprehension is

HOW WELL DO ADULTS READ? 69

measured by the ability to get the general idea of the content.
Controlling the construction of the tests prevented ascribing the
difference in mean scores to a difference in difficulty of the tests
themselves. Caution had been taken to select items for one test-
form that would closely parallel those of the other when ranked
by the occurrence of elements of expression believed to be indi-
cators of difficulty. Furthermore, the highest possible scores on
the two forms were approximately equal, being 92.2 for fiction
and 91.4 for non-fiction. It seems reasonable to infer, therefore,
that fiction is easier than non-fiction, though the two types of
material contain a similar number of elements of difficulty.

That adults varied more widely in their understanding of fic-
tion than non-fiction is evident from the standard deviations of
the two tests. A difference of 2.77 .56 between the units of
variability, 24.01 and 21.24, is statistically significant.

As yet, it has not been possible to determine the level of
achievement which persons should attain on the Adult Reading
Test. If we assume that they should make a score of 50 or more
in other words, that they should comprehend at least half of
what they read as measured by the tests used then it is clear
that somewhat more than half the group, 60 per cent, inter-
preted fiction satisfactorily. A slightly smaller number, 56 per
cent of the group, interpreted non-fiction satisfactorily. If we
assume that a score of 40 or more is satisfactory, then approxi-
mately 30 per cent showed a deficiency in comprehension of
general materials.

Further comparison of the ability of adults to read the two
types of material is afforded by Table XVI, which presents the
mean scores on the fiction and non-fiction tests obtained by the
twelve groups of adults. The mean scores on the fiction test
varied from 19.95 for the A and M Junior High School group to
75.40 for the parent-teacher group. According to these data, the
latter group was on the average about four times as efficient in
reading fiction as the former. The mean scores of at least four
groups fell below what is probably the minimal desirable score
of 40.

WHAT MAKES A BOOK READABLE

An examination of actual scores made by individual subjects
shows that 89 per cent of the A and M Junior High School group
made scores below 40, whereas only 2.4 per cent of the parent-
teacher group scored as low as 40. Scores of the same two groups
defined the range on the non-fiction tests. Again the A and M
Junior High School group attained the lowest mean score, 19.85;
and the parent-teacher group the highest mean score, 67.45*

TABLE XVI
SCORES FOR EACH OF TWELVE GROUPS ON THE ADULT READING TESTS

Group

Mean Score on
Fiction Test

Mean Score on
Non-Fiction Test

Parent-teachers

7C.4O+ .8^

67.4.<4-I 12

Berea Academy

66.78i.o5

c8.oo+ .84

Berea College

66 274- .Q<

60 87 88

Beny College

64.60+1.30

<7 8<i .10

Rabun Gap

62.204-1.90

r?.qr-fi .72

Betty High School* (Juniors and Seniors)

cS.io

CO OO

Berry High School (Freshmen)

4.6 . 77 -f 2 . 1 T

^0. 72 4- 2. 1O

A & M Senior High School

40.26 + 1.43

30.474-1 .02

Evening schools

70 . 62 + 1 . QO

-3Q 2AI .A.6

Berea Foundation

T7.I7I .<7

T7.OC+ 1 ,7O

Glenn Street*

^6.70

28.70

A & M Junior High School

IQ.CX-tl .4.1

19 85i 26

Entire group

ro 86 . <Q

47 864* C2

T/-""- 1 - O*

* Probable errors not computed because of small number of cases.

For six of the twelve groups, the average score fell below 40.
That all groups tested were able to read fiction more effectively
than non-fiction is shown further by Table XVI. In every case,
the mean score obtained on Form I was higher than on Form i.
A more concrete picture of how well adults read the two types
of material is presented in Figure 6. Comparison can be made on
the basis of three values: median, third quartile, and first quar-
tile. The median marks the point above and below which half
the scores fall. The third quartile designates the point in the
distribution above which one-fourth and below which three-
fourths of the scores fall. The first quartile indicates the point

I 1

v/s/x

KXX/V

_..,-,

tV ///?"s/s_/x/s/ //////A

4)
fl

- 1

' I

= .1

72 WHAT MAKES A BOOK READABLE

above which three-fourths and below which one-fourth of the
scores fall. If the first and third quartile are widely separated,
it is evident that there was a marked variability within a group.
If, on the other hand, the quartiles are close together, it is plain
that the scores within the group were less scattered.

Certain tendencies are apparent from Figure 6. With but one
exception, all groups made a higher median score on fiction than
non-fiction. In all but one case, the third quartile of the former
was conspicuously higher than that of the latter. A comparison
of the inter-quartile range for the two tests supports the evi-
dence indicated by the standard deviations in Table XV, name-
ly, that the ability of the adults tested varied more in read-
ing fiction than in reading non-fiction. There is some evidence
that groups classified at higher educational levels read both
types of material with greater understanding than groups classi-
fied at lower levels, and that the spread of ability tends to be
markedly less. That parent-teacher groups read with more un-
derstanding and less variability is to be expected. Since they
are non-school groups, it is probable that they read general
reading-matter more than do groups who attend school. Fur-
thermore, inasmuch as they volunteered to take the tests, one
may assume that they represented the most able readers. Cir-
cumstances prevented other groups from being as selective.

In Table XVII we have compared ability to read fiction and
non-fiction in terms of a single value, the coefficient of correla-
tion, 2 If a particular group reads fiction and non-fiction equally
well, as measured by the tests used, the correlation will be high
and the coefficient will approach +i. If a group tends to read
one type of material better than the other, the correlation is
low and the coefficient will approach zero.

The greatest agreement in reading fiction and non-fiction is
shown in Table XVII by a coefficient of .779 for the evening-
school groups. The least agreement is .481 for the A and M

2 Statistical comparisons were made by Margaret D. Cleary, "Achievement in
Reading of Selected Groups of Adults." Unpublished Master's thesis, Department
of Education, University of Chicago, 1933, p* 84.

HOW WELL DO ADULTS READ? 73

Junior High School group. Even here, however, the correlation
is significant. Comparisons between the correlations obtained
for different groups are not warranted, for the reason that the
groups are not comparable in age, race, educational level, or
socio-economic background. The general conclusion to be drawn
from this table is that the relationship between ability to read
fiction and non-fiction is positive and significant, and in most
cases fairly high.

TABLE XVII

COEFFICIENTS OF CORRELATION BETWEEN FORM i AND
FORM 2 OF THE ADULT READING TEST

Groups

r P.E.

Evening schools

.770+ cm

Berea Foundation

.778+ 028

Rabun Gap

7T7 + .048

Berry High School (Juniors and Seniors)*

7OO

Berry High School (Freshmen)

.601+ oci

Parent- teachers

.666 4-. 041

Berea College . ...

66c+ 0^8

A & M Senior High School

, <o6H- cxi

Glenn Street School* .

O7 WJ - - W 3*
.<67

Berry College

. <7Q~r- .OCO

Berea Academy

. <IQ .O<O

A & M Junior High School

.481 + .07^

* Probable errors not computed because of small number of cases.

MEASURING THE READING ABILITY OF ADULTS IN
TERMS OF GRADE NORMS

The current interest in improving the reading habits of adults
has led to numerous conjectures relative to the grade levels rep-
resented by the reading ability of adults. In the previous sec-
tions we have summarized important facts relative to the ability
of adults to comprehend the content of general books, maga-
zines, and newspapers. The most serious limitation to a thor-
oughgoing interpretation of these facts is the lack of standards.
What the various scores mean in relation to graded progress
must be determined by more extensive experimentation.

74 WHAT MAKES A BOOK READABLE

To supplement the data already presented, it was thought
advisable to measure reading achievement in terms of grade
norms. From these could be ascertained within certain limita-
tions, first, what is the general grade level of adult reading
ability; and, second, whether ability is markedly different
among different classes of adult readers.

In the absence of standardized adult tests to serve this pur-
pose, it was necessary to utilize a general test which carries a
wide range of grade norms. The one chosen was the Monroe
Standardized Silent Reading Test. It consists of a series of
short passages followed by questions or directions. In some
cases the question is answered by drawing a line under the word
representing the right answer. In other cases, the examinee is
directed to write out the answer to a question in order to show
his understanding of the item. The tests have been prepared to
measure a wide range of reading achievement: Test I, for use
in grades III, IV, and V; Test II, in grades VI, VII, and VIII;
and Test III, in grades IX to XII, inclusive. The number of
exercises answered correctly in a period of four or five minutes
makes the score. The raw scores can be transmuted into "B"
scores, ranging from 2.0 to 17.0+- They indicate the school
grade in which pupils normally attain the equivalent raw scores.

ADDITIONAL GROUPS TESTED

In addition to the 756 adults comprising the groups shown in
Table XI, certain other groups were given the Monroe test.
They are listed in Table XVIII. These groups made a total of
1,690 adults whose achievement in silent reading was deter-
mined in grade norms. Probably the most important and un-
doubtedly the most representative group numbered 151 adults,
of whom 75 lived in Coatsburg, Illinois, and 76 on adjacent
farms. The village people included all men and women of the
total population of 200 who were willing to co-operate. A ran-
dom sampling was taken from the surrounding rural area. The
mean age of the village group was 44,9 years, and of the rural
group, 43.5 years. In each case, the mean grade last attended

HOW WELL DO ADULTS READ?

in school was the eighth, the extremes ranging from third to
twelfth in the village, and from fourth to twelfth on the farms.
Occupational interests found in Coatsburg are representative of
a typical village. In professional service were the doctor and
banker; in trade, the merchant; in building and construction,
the carpenter, painter, and paper-hanger; in mechanical indus-
tries, the automobile mechanic; in agriculture, the farmers; and
in the unskilled labor group, the majority of the citizens who

TABLE XVIII

ADDITIONAL GROUPS TESTED ON THE MONROE STANDARDIZED
SILENT READING TEST

Name

Grade-
Place-
ment

Geographical
Location

Number

Cross-section of rural popula-
tion
i . Villagers

Coatsburg, 111.

2. Farmers

Adjacent area

Sampling of negro population
i* Indigent adults . . . .

Columbus, Ohio

IQ.I

2. Selected negro men

Chicago, 111.

I2<

3. Florida group

1 8 Florida cities

<?V7

Total

Q^A

work at what they can when they can. The reading ability of
these persons is probably highly representative of the reading
ability of similar populations in the Middle West.

Because a large proportion of the negro population has had
limited educational opportunities and engages in little reading,
it seemed desirable to study the reading achievement of several
groups in widely distributed areas. One group included iai in-
digent men and women who applied for aid at the Goodman
Guild in Columbus, Ohio, 3 and who stated that they were able
to read and write. Of this number, 48 were women and 73, men.

3 Data relative to this group were obtained by Dale and Tyler in their study of fac-
tors of difficulty inherent in materials pertaining to the field of health.

76 WHAT MAKES A BOOK READABLE

The mean age for the entire group was 26.2 years. The previous
education claimed by this group varied from little or no school-
ing to high-school graduation. The mean grade last attended
was the seventh.

A second group included 125 negro men of Chicago. 4 Because
of the greater ease with which group contacts could be made,
subjects were selected from such organizations as the lodge of
Elks and Legion posts. Of the 265 who were asked to co-operate,
140 refused. The most common reason for refusal, given by
22.5 per cent, was inability to read. The rest of the 140 were
less frank, but their reasons suggested poor reading ability. It
is evident, therefore, that the 125 who co-operated in the study
were not altogether representative of the negro male population
of Chicago. The amount of education reported by this group
ranged from fourth grade to third year in dental college, the
median falling at eighth grade. An occupational classification
of this group showed that the largest number, 20 per cent, were
postal employees. The remainder were engaged in minor types
of service ordinarily rendered by negroes in a large city.

A third group included 537 colored adults in eighteen cities
of Florida. All were engaged in group educational activities un-
der the direction of the Agricultural and Mechanical Arts Col-
lege at Tallahassee, Florida. Their mean age was 38 years. The
mean grade last attended in school was the ninth. Clearly this
group, like the Chicago group, was composed of negroes some-
what above the average of their race in formal education,

AT WHAT GRADE LEVELS DO ADULTS READ?

The achievement records of all groups on the Monroe test
are summarized in Table XIX. The left-hand column indicates
the range of "B" scores, or grade scores. The columns to the right
show the number and percentage of adults making each score.
The entries in the table show, first, a very wide range of scores.
This is to be expected, since the groups differed markedly in

4 This part of the study was carried out by G. T. Wiggins, whose complete report of
'The Reading Habits, Interests, and Achievements of Negro Male Adults" is in un-
published form in the Department of Education Library, University of Chicago.

HOW WELL DO ADULTS READ?

educational status. Seven adults obtained a score of 16.95 or
above, that is, approximately grade 17. Seventy-seven received
scores below 2.95, or less than the norm for grade 3. The mean
score, 7.81, shows that the 1,690 adults tested were able to read
with an average proficiency equal to the normal expectation of

TABLE XIX

NUMBER AND PERCENTAGE OF ADULTS IN SIXTEEN REP-
RESENTATIVE GROUPS MAKING VARIOUS GRADE SCORES
ON THE MONROE STANDARDIZED SILENT READING TEST

Range of "B" Scores

Number

Per Cent

16 . 95 or above

O 4.1

IC.QC l6.QC

O. C7

14. QC I C .QC

I 78

17. Q< IA.QC

I O2

6 OA

12. QC IT.QC

122.

7.22

II .QC I2.QC

3.67

IO QC II.QC

A QI

9.QC IO.QC

4 ^8

'yj AW> :O
8 QC- Q.QC

184.

IO.SQ

7 QC 8 Q<

4.26

6 QC 7 QC

4.8C

C .QC- 6 QC

170

10. 06

A QC C QC

I2C

7.40

Q C 4 Q C

24.^

14.38

2.QC- 1 QC

2 4 8

14.66

-yi o-y:>
Less than 2 95

4 .c6

Total

1600

IOO.OO

Mean 7.81+ .06
S.D. =3.891.04

pupils in the eighth month of the seventh grade. Somewhat less
than half of the adults tested, 44 per cent, had reached or sur-
passed a level of achievement equivalent to the reading norm
for pupils in the eighth grade of the elementary school.

Further inspection of the entries in the table shows that the
scores tended to cluster around certain grade levels, namely,
third, fourth, fifth, sixth, ninth, and thirteenth. The fact that
about one-third of the persons tested ranked below fifth grade

78 WHAT MAKES A BOOK READABLE

on the Monroe test recalls an observation made earlier with
respect to achievement on the adult reading tests. One-third of
the persons whose reading ability was measured by the latter
tests were found to obtain scores below 40 out of a possible 92.
It seems significant that the same proportion of both groups
should manifest an ability in reading that undoubtedly pre-
vents them from engaging satisfactorily in mature reading ac-
tivities. Since the intelligent functioning of a democratic social
order implies an intelligent citizenry, it seems socially impera-
tive that the reading efficiency of large elements of our popula-
tion should be improved or that more materials suited to their
level of achievement should be prepared. The first proposal is
receiving increasing attention from civic and federal authorities
who have combined forces in the interest of adult education. In
connection with the alternative proposal, there is a persistent
need for more materials appropriately graded for adults whose
reading scores fall below 6.95.

GRADE LEVELS ATTAINED BY TWELVE GROUPS
OF 756 ADULTS

Table XX shows "B" scores made on the Monroe test by the
twelve groups described earlier. The mean score, 9.68, is sig-
nificantly higher than the mean score for all groups tested. De-
spite this fact, about 30 per cent of the group failed to attain a
level of efficiency which approaches mature reading, that is,
seventh grade. Again, the observation can be made that ap-
proximately the same percentage fell below a score of 40 on the
Adult Reading Test.

Figure 7 has been prepared to show more clearly the distri-
bution of scores obtained by the twelve groups. As can readily
be observedj the scores tend to cluster around three areas.
About one-third fall in the area from 2 to 6; another third range
from 7 to ia; and the remainder, from 13 to 17. These modal
tendencies mark roughly the elementary, secondary, and col-
lege levels. Were the measurement of reading ability to be ex-
tended to include a wider sampling of adults, the distribution of

HOW WELL DO ADULTS READ?

obtained scores would presumably conform fairly well to the
normal curve. It is apparent from the evidence shown in Fig-
ure 7 that adults are substantially differentiated with respect
to reading ability.

TABLE XX

READING ABILITY OF 756 ADULT SUBJECTS AS MEAS-
URED BY THE MONROE STANDARDIZED
SILENT READING TEST

Range of "B" Scores

Number

Per Cent

1 6 95 and over

o 8

IC.QC 16 QC

0.8

I4..QC If. OC

3.6

10 or i A QC

II 2

12 cK 11 Q<

J
07

12 8

II OC 12. Q<

<.A

IO.QC II .QC

4-4.

c.8

9 or 10 Q< ...

6 8

yj lv- 7j
8 Q C Q . Q <

12.6

7 . Q C 8 . Q C

t.'i

6.QC 7.QC

4.2

C QC 6 Q<

8.2

D -iO W> :O

A.QC C.Q<

6.8

1.QC 4..QC

9.0

2 QC 7 Q<

r 8

I .QC 2Q<

O.Q

Total

7C6

IOO.O

The mean achievement of each group and the standard devia-
tion of the mean are given in Table XXL In general, there is a
definite tendency for the mean score to increase with the num-
ber of years in school. Certain exceptions may be noted. For
example, both the Berea Foundation and the A and M Junior
High School groups fall below the Glenn Street School group,
which was classified as elementary. This circumstance may be
explained in part by the fact that the two junior high schools in
question admit young people of limited education. The parent-
teacher groups rank somewhat lower than their educational

WHAT MAKES A BOOK READABLE

status leads one to expect, and substantially lower than seems
compatible with their superior accomplishment on the adult
tests. It seems reasonable to attribute the rank on the Monroe
test to the time element which tends to penalize adults who are

Per Cent

'95 '-95 3-95 4-9S 5-95 6 -9J 7-9S

9-95 '95 "95 12 -95 U-9J

' 6 V5 '795

FIG. 7. Scores of 756 adults on the Monroe Standardized Silent Reading Test

not attending school. The high rank of the Berry High School
group probably has little real significance because of the small
number of cases. In spite of these exceptions, the data imply
that reading efficiency continues to improve with attendance at
school, even at upper levels. Although the differences in mean
scores at successive levels may not be highly significant, the
general trend of improvement cannot be dismissed lightly.

HOW WELL DO ADULTS READ?

For the sake of further comparison, let us assume that adults
whose reading ability measures 7.0 or above have attained a
reasonable proficiency in reading. We may then ask, What per-
centage of adults in each group failed to reach the acceptable
level of seventh-grade efficiency? The answer to the question

TABLE XXI

MEANS AND STANDARD DEVIATIONS OF THE SCORES OF TWELVE

ADULT GROUPS ON THE MONROE STANDARDIZED

SILENT READING TEST

Group

Mean

S.D.

Berry High School (Juniors and Seniors)*. .
Berea College

14.10
I2.56+.I8

2.04
2.68.i3

Berry College

12 94. 4- IQ

2 7O+ .14.

A & M Model Senior High School

II .6l+ .27

3.48+ .IQ

Berea Academy

II 34 + . 2O

2.89 .14

Rabun Gap

10. 2O + .W

7.17+ .27

Parent-teachers ...

Q 47i I<

I 96+ IO

Berry High School (Freshmen) . .

j'*rl -L *.)
7 QC + A.1

4.l6+ .70

Glenn Street School*

6 co

2.7C

A & M Model Junior High School

6 . 24 + . 28

2. 86 .20

Berea Foundation

6. 09 .14

2.OI .IO

Evening schools .

C 40 i . 14

1 .70+ .IO

Entire group

Q.68+ .00

3.82+ .06

* Probable errors not computed because of small number of cases.

is shown in Figure 8. Evening schools were found to contain the
highest percentage of adults with scores below 7.0, namely, 8.2.
None of the Berry High School group attained scores below 7.0.
As explained earlier, they were too few in number for reliable
generalizations.

An important fact revealed by the diagram is that a large
number of deficient readers were found in most of the groups
tested. This circumstance can hardly be attributed to mere
chance. It seems rather to suggest that deficient adult readers
persist somewhat generally. Abundant evidence is available to
support the present findings concerning the inadequate prepa-
ration of high-school and college students to read much of the

8* WHAT MAKES A BOOK READABLE

material prescribed for them. Less direct but equally convinc-
ing evidence of 'immature reading ability among adults may be
gleaned from facts pertaining to their reading interests and
habits.

Evening schools
Berea Foundation
A & M Junior High

Glenn Street School

Berry High School
(Freshmen)

Rabun Gap
Berea Academy
A & M Senior High
Parent-teachers
Berry College
Berea College

Berry High School
(Juniors and Seniors)

Per cent o

30 40 50 60 70 80 90

FIG. 8. Percentage of adults below seventh-grade level of reading achievement
as determined by the Monroe test. (Adapted from Geary, op. cit., p. 50.)

ACHIEVEMENT ON MONROE TEST COMPARED WITH
ACHIEVEMENT ON THE ADULT READING TESTS

A comparison of Table XVI with Table XIX reveals dif-
ferences in the achievement of 756 adults as measured by the
adult reading tests and the Monroe test. In some cases, the
rank attained in reading adult materials is lower than the grade
rank in comprehension seems to warrant. Four conspicuous ex-
amples are the Berry High School Juniors and Seniors, who
ranked first on the Monroe test and sixth on the adult tests;
the A and M Seniors, who ranked fourth and eighth respective-
ly; the Glenn Street Elementary students, whose rank of nine

HOW WELL DO ADULTS READ? 83

on the Monroe test was lowered to eleven when reading adult
material; and the A and M Juniors, whose grade rank in silent
reading was tenth and whose rank in reading adult material was
twelfth and lowest.

More striking variations may be noted in the opposite direc-
tion for groups who read adult material relatively better than
their grade-rank in silent reading implies. For example, parent-
teacher groups ranked first on the adult tests and seventh on
the Monroe test. Berea Academy, given second place in read-
ing materials commonly read by adults, ranked fifth according
to grade norms. The evening-school groups showed a difference
in rank from nine to twelve on the respective measures of read-
ing ability.

A rough indication of the relationship between ability to read
as measured by the two tests is shown by a rank correlation of
. 623.5 More specific relationships are given in Table XXII.
These relationships are expressed in terms of coefficients of cor-
relation which indicate the agreement between achievement on
the Monroe test and comprehension of materials in Forms i
and 2 of the adult test. None of the coefficients is high enough
to show close agreement. The disparity in relationship may be
due to several influences. The tests may not be altogether re-
liable, or they may measure somewhat different abilities. Some
groups may read adult materials with a proficiency beyond that
manifested by the Monroe test because of familiarity with gen-
eral materials. The explanation may He in the timed conditions
attached to the latter test or in some other contributing factor.

Probably the most significant fact to be derived from the
table is that in general the scores of adults ranking lowest on the
Monroe test correlated highest with scores on the adult tests.
One may infer, therefore, that for adults of limited proficiency

s Spearman's formula based on rank differences is

Karl J. Holzinger, Statistical Methods for Students In Education (Boston: Ginn & Co.,
1928), p. a78.

8 4

WHAT MAKES A BOOK READABLE

in reading the Monroe test indicates with reasonable accuracy
the ability of the group to read adult fiction and non-fiction of a
general sort.

DO ADULTS TEND TO ATTAIN HIGHER LEVELS OF COMPRE-
HENSION WITH INCREASED EDUCATIONAL TRAINING?

This question can be answered in a general way by comparing
the mean reading achievement of each group, Tables XX and

TABLE XXII

CORRELATIONS BETWEEN SCORES ON FORMS i AND 2 OF THE ADULT

READING TEST AND THE MONROE STANDARDIZED

SILENT READING TEST

Groups

Monroe and
Fiction
(Form i)

Monroe and
Non-Fiction
(Form 2)

Glenn Street*

.61

Rabun Gap

, CQ + .07

.65.o6

Parent- teachers

50 + 06

4.C + 06

Berea. Foundation

A,Q + ,OC

.4/1+ .06

A & M Senior High School

48+ 06

27+ O7

Evening schools

.4.1+ O7

.31 08

Berry High School (Freshmen)

4! 08

?< + OQ

A & M Junior High School

38 08

18+ 10

Berry College

. *u .06

. i r + . 06

Berry High School (Juniors and Seniors) * . .
Berea College ....

3o .06

25 + 06

Berea Academy

28+ 06

42 06

* Probable errors not computed because of small number of cases.

XXI, with educational training, Table XL There is evidence
of close agreement between educational training and mean
achievement on the Monroe test, the most obvious exceptions
being noted in the Berea Foundation, the A and M Junior High
School, and the parent-teacher groups. Less agreement as
shown by rank order is found between educational training and
ability to read fiction, as measured by Form i of the adult test.
While there is no reason to expect absolute agreement, some
variations are both unexpected and interesting. The apparent
superiority of the Berea Academy to other secondary school

HOW WELL DO ADULTS READ?

groups on the Monroe test is even more marked than on the
fiction test, where it ranks slightly above the college groups
tested. Satisfactory explanations for this circumstance cannot
be found in the evidence at hand. The A and M Junior High
School ranks not only lower than other groups similarly classi-

Glenn Street School

Evening schools

Berea Foundation

A & M Junior High

Berea Academy

Berry High School (Freshmen)

Rabun Gap

A & M Senior High

Berry High School
(Juniors and Seniors)

Berea College

Berry College

II III IV V VI VII VIII IX X XI XII XIII XIV XV X\

FIG. 9. School grade and reading grade achievement. (Portion in black represents
reading retardation; shaded portion, reading acceleration.) (Adapted from Cleary,
op. cit., p. 47.)

fied in ability to read fiction but lower than groups classified as
elementary. Similarly, the A and M Senior High School group
ranks conspicuously lower than others with which it may rea-
sonably be compared. These two groups, representing elements
of the negro population, are handicapped by meager reading
facilities. They probably engage in little independent reading.
In Figure 9 is given a graphic comparison of reading achieve-
ment, as measured by the Monroe test, and general educational

86 WHAT MAKES A BOOK READABLE

training, as indicated by grade-placement or last grade in school.
While the figure shows plainly that reading achievement tends
to parallel educational training in a general way, only four of
the eleven groups had acquired a reading ability at or above the
expected norm. The Glenn Street School and the A and M Senior
High School groups reached average expectancy for their levels
of schooling, while the Berry High School (junior and senior)
and Berea Academy groups exceeded it. The tendency for read-
ing retardation to grow less in amount as schooling progresses
is to be expected, since disability in reading would naturally be
reflected in a lack of general progress. Exceptions are to be
noted in the two college groups. Since the parent-teacher group
was not in school, no attempt was made to estimate their ratio
of achievement.

Despite the fact that there is a general correspondence be-
tween average reading achievement and educational training,
individual records tend to depart from this general trend.

ACHIEVEMENT OF THE VILLAGE GROUP

What is the reading ability of adults living in village and rural
areas of the sort represented by Coatsburg, Illinois? This ques-
tion can be answered by reference to Table XXIII. Entries in
the table show that mean score, range in achievement, and vari-
ability of the group scores are similar for the village and rural
groups. The mean score of the combined group, 7*92, agrees
closely with the mean score of the total 1,690 adults tested,
shown in Table XIX to be 7.81. In view of the fact that the
total group included college students and college graduates, the
similarity in mean achievement is not expected. This agreement
suggests that the Coatsburg groups may fairly represent the
class of reader termed "average" in larger and more diversified
groups. The investigation needs to be extended, however, be-
fore we can define the reading level of the average adult at ap-
proximately eighth grade, as the present data indicate.

What is the reading ability of adults when they read orally?
This question arose in connection with the Coatsburg groups.

HOW WELL DO ADULTS READ?

From their mean ages, 44.9 and 43.5 years, the inference seemed
warranted that their training in reading might have been re-
ceived at a time when oral rather than silent reading prevailed
in classrooms. Their proficiency in oral reading, therefore,
might presumably surpass that of silent reading.

TABLE XXIII

NUMBER AND PERCENTAGE OF 151 VILLAGE AND RURAL ADULTS MAKING VARIOUS
GRADE SCORES ON THE MONROE STANDARDIZED SILENT READING TEST

GRADE SCORE

75 VILLAGE
PEOPLE

76 FARM ADULTS

COMBINED
GROUP

No.

Per Cent

No.

Per Cent

No.

Per Cent

1 6 . 95 or above

1.32
2.63

2
2
4

7
7

J 3
8

IO
IO
12
*3

0.66
1.32
1.32
2.65
4.64
4.64
8.61
5-30
8.61
6.62
6.62
7*95
*5- 2 3
16.56
8.61
0.66

IC.QC l6.QC

I A QC 1C QC

i
i

3
5
7
9
16
8

2.67

i-33
6.67
8,00
10.67
4.00
2.67
4.00
6.67
9-33

12. OO

"33

10.67

17 .QC 14. QC . ....

5
5
ii

7
5
5
14
9
5
i

3-95
2.63
1.32
6.58
6.58
14,47
9,21
6.58
6.58
18.42
11.84
6.58
1.32

12. QC 17. QC

JJ A J .7J
II .QC 12. QC

IO.QC II -QC

9QC IO QC . ...

iO xvy ';O
8 QC Q.QC

7 QC 8.QC

6 QC 7.QC

C.QC 6.QC. . .

A QC C QC

*r'7j j'yj
3QC A QC

2.QC- 7.QC

Less than 2 . 95

Totals

IOO.OI

15*

100.00

Mean

7.80+ .28
3.6i.2o

8.o 4 .27
3-4j-*9

7.92.i 9
3-53-*3

S.D

Interesting comparisons are afforded by the data in Table
XXIV. The table shows the last grade in school attended by
members of the village and rural groups and their grade scores
in oral and silent reading. The oral reading scores were ob-
tained through the use of the Standardized Oral Reading Para-
graphs, which measure achievement in terms of rate of oral

WHAT MAKES A BOOK READABLE

reading and number of errors made. The average grade score
in oral reading for the village adults is 9.1 ; in silent reading, 7.8.
For rural adults these scores are increased respectively to 9.7
and 8,0. Both groups show an average achievement in oral read-

TABLE XXIV

GRADE LAST ATTENDED AND GRADE SCORES IN ORAL READING AND IN COMPRE-
HENSION FOR 75 VILLAGE AND 76 RURAL ADULTS

GRADE

75 VILLAGE ADULTS

76 RURAL ADULTS

Grade
Last

Attended

Oral
Reading
Grade
Score

Compre-
hension
Grade
Score

Grade
Last
Attended

Oral

Reading
Grade
Score

Compre-
hension
Grade
Score

3
5
3
32
8
8

7
8
o
o

i
o
i
4
3
4
II

9
13
27

o
o

5
9
14
5
5
7
ii

5
5
i

2
I

III

4
6

*3
8

7
3
5
ii

12
II
20
2

I
O

8
16
9
7
5
3

3
8
6

5
i

VII

VIII

XII

XIII .

XIV

XVI

XVII

Total. ..
Mean. . .

9-i

7.8

9-7

3.04

ing approximately one and one-half grades in advance of aver-
age achievement in silent reading. Since the mean grade last
attended by both groups was the eighth, it is apparent that
adults in this community were sufficiently well trained in the
mechanics of reading to surpass, years later, the norm for that
grade* Their mean achievement in silent reading corresponds

WHAT MAKES A BOOK READABLE

more closely to the amount of schooling received than was
noted with respect to the twelve groups reported earlier.

The difference in achievement in oral and silent reading is
strikingly illustrated by Figure 10. While it is evident that more
adults received high scores in oral than in silent reading, it is

TABLE XXV

NUMBER AND PERCENTAGE OF ADULTS IN EACH OF THREE NEGRO GROUPS
MAKING VARIOUS SCORES ON THE MONROE STANDARDIZED
SILENT READING TEST

GRADE SCORES

COLUMBUS
GROUP

CHICAGO
GROUP

FLORIDA
GROUP

TOTAL

No.

Per
Cent

No.

Per

Cent

No.

Per

Cent

No.

Per

Cent

tr of 16 oc

i
i

8
9

2
. . 4

0.80
0.80
6.40
7.20
i. 60
3.20
4.80
14.40
4.80

12. CO
18.40
8.80

8.00
8.80

*3
18

14
26

IS
76

22
4
9 6

51
150
I 9 I
69

0.13
0.13
1.66
2.30
1.79

3-32
i ,92
9.71
2.81
5.11
12.26
6. 51

I9.l6
2 4-39

8.81

A .) -yj AVJ *!?J
Hnf ir or

I3.QC Ii.Q<

5
9
ii

5 2
14

22
56
32

118
146
42

o-93
1.68
2.05
4.10
1.49
9.68
2.61
4.10

10.43
5.96
21.97
27.18
7.82

I2.Q<~ IT.oC

II .QC I2.Q<

0.83

IO QC II <X

9QC 10 <K

i
6

8
22

0.83
4.96
1.65
2.48
14.05
6,61
18.18
28.10
22.31

6
18
6
15
^3
ii

8.CX Q.CX

7.<X- 8.Q*

6.<K 7.QC

<.Q< 6.oc

4..q< C.QC

T.CK 4..QC

2.<K~ ^.Q<

-yj o-io
Less than 2 gt

Total.

121

JOO.OO

125

100.00

537

100.00

783

ICO.OI

Mean

4.57-*3
2.o8 .09

8.20 .20

3-3i-*4

5*79-9
2.95.o6

5. 9 8.o7
3-^-5

S.D

quite as evident that some adults ranked low on both types of
reading. It is doubtful, therefore, if they can read and compre-
hend satisfactorily much of the adult material now published
that should be of interest to them. Probably at least 40 per cent
of the village and rural groups are able to read only relatively
simple reading matter.

HOW WELL DO ADULTS READ? 91

HOW WELL DO NEGRO ADULTS READ?

Three groups of negroes numbering more than 800 were in-
cluded in the study. The distribution of their scores is given in
Table XXV. A cursory glance shows that most of the negroes
ranked low in comprehension. The mean of the Columbus group
is 4-57- This is the achievement normally expected of pupils
in the fifth month of the fourth grade. The mean of the Chicago
group is 8.2,0; of the Florida groups, 5.79; and of all groups,
5.98. At the lower extreme of the range it may be noted that
below the fourth grade norm are 50 per cent of the Columbus
group, about 10 per cent of the Chicago group, and 35 per cent
of the Florida group. Only if per cent of the Columbus group
rank as high as tenth grade in reading achievement, whereas
25 per cent of the Chicago group reach or exceed that level.

It may be remembered that the group tested in Columbus was
composed of indigent negroes of little education. The Chicago
group, on the other hand, included employed negro men, whose
formal education had extended on the average to eighth grade.
Significant differences in the mean achievement of these groups
probably may be explained, therefore, by the selectiveness of
the groups. If we assume that a "B" score of 7.0 that is, a level
of reading achievement normally attained in seventh grade is
desirable, then it is evident from Table XXV that approximate-
ly 70 per cent of the negroes tested are unable to engage in-
telligently in reading activities at the adult level.

CONCLUDING STATEMENT

From the findings presented in this chapter, it is obviously
impossible to arrive at any very far-reaching conclusions con-
cerning the reading efficiency of adults. The question of how
well adults read must be answered ultimately in terms of par-
ticular groups reading different kinds of materials for a variety
of purposes. Such an answer cannot be obtained from our find-
ings, for the reason that they lack the specificity implied in the
foregoing statement. The number of adults tested is too few to
permit classification as to age, sex, race, occupation, educational

92 WHAT MAKES A BOOK READABLE

background, reading experiences, and other factors which pre-
sumably influence reading ability. The test materials, although
restricted to what we are wont to term "general" materials,
deal with a variety of subjects seemingly representative of
varying degrees of interest and familiarity. The outcomes of
reading by which ability is measured fall far short of all the
outcomes of adult reading activities.

It is clear from these limitations that we have made no more
than a broadside attack on the problem of how well adults read.
If the findings obtained in this fashion have even a small degree
of reliability, we may expect to find from extended studies in-
formation concerning reading ability that will have practical
value for workers with adult groups.

The specific conclusions and implications warranted by the
findings presented in this chapter may be summarized as fol-
lows:

I. About one-sixth of the adults tested are able to read with
a degree of proficiency normally achieved by high-school gradu-
ates. Half of this number have attained the norm of college
students. Only one-half of one per cent, however, were found to
read with the proficiency expected of college graduates. It seems
reasonable to conclude, therefore, that adults in this upper one-
sixth of the total sampling are sufficiently skilled to read the
more difficult types of materials now available.

a. Approximately one-half of the adults tested have attained
a sufficiently high level of achievement to enable them to read
with reasonable ease and understanding most of the general ma-
terials now prepared for adults. For this group, radical changes
in the character of available material do not seem imperative.
However, a wider production of high-grade material, slightly
less difficult than the "better" fiction and non-fiction now pro-
duced, should be of distinct advantage.

3. At least one- third of the 1,690 adults who co-operated in
the study have not acquired sufficient skill to enable them to
engage intelligently in adult reading activities. They read gen-
eral materials in books, magazines, and newspapers with an

HOW WELL DO ADULTS READ? 93

average score of 40 out of a possible 90. Their achievement in
silent reading, as measured in terms of grade norms, does not
exceed fifth grade. For them the enriching values of reading are
denied, unless materials reflecting adult interests be adapted
to meet their needs. The data in this chapter suggest that the
materials provided should correspond in difficulty to those ap-
propriate for use in the fourth, fifth, and sixth grades. Such a
proposal presupposes the availability of still simpler materials
for use in promoting functioning literacy and in establishing
fundamental reading habits.

4. A considerable portion of the adults tested,, approximately
one-sixth, is able to read only the simplest newspapers, maga-
zines, and books. A smaller percentage, one twenty-fifth, is able
to read only such materials as might appropriately be used in
the second grade of the elementary school. Reading content
prepared for primary children falls far short of meeting adult
needs, yet simple adult material of a high order is not readily
available. If the reading efficiency of this group is to be im-
proved through definite training, what materials, then, should
be used? The problem is vital and timely. It seems challenging
enough and sufficiently widespread in its application to warrant
the intensive study of the difficulty of reading materials as pre-
sented in the remaining chapters of this report.

To answer precisely the question of how well adults read,
there is evident need for a series of carefully controlled studies
utilizing various groups of adults, each of which is approximate-
ly alike in those personal characteristics that presumably in-
fluence reading ability. Such studies will aim to define specific
reading abilities for each group with respect to the kind of ma-
terial read and the outcome desired.

CHAPTER IV

WHAT ELEMENTS INFLUENCE THE DIFFICULTY
OF ADULT READING MATERIALS?

THE preliminary study reported in chapter iii sought to
evaluate the reading efficiency of adults who were clas-
sified at various educational levels. The evidence pre-
sented indicates that in practically every group tested there was
a large percentage of adults who had not attained the maturity in
reading expected of them. It seems probable that the reading
ability of adults in general has been greatly overestimated by
persons who prepare materials for adult readers. It seems prob-
able, also, that much of the reading matter now available is too
difficult for readers of limited ability to understand. Hence it
fails to reach many persons who are interested in its content and
who would read it if they could. Without question, some of this
material would reach a larger audience were a reliable technique
devised by means of which difficulty could be determined with
accuracy.

In the present chapter two assumptions have been made. One
is that there are elements inherent in reading materials which
are significant indicators of difficulty; the other, that the identi-
fication of these elements is an essential step in developing a
technique which will help to solve the problem of meeting the
individual reading needs of adults. The purpose of the chapter
is, first, to identify the elements of difficulty which we have as-
sumed are inherent in reading materials; and, second, to find
out how these elements may be utilized in a more accurate esti-
mate of the difficulty of specific books, magazines, and news-
papers.

As stated in chapter i, we have limited our study to a determi-
nation of elements of difficulty inherent in reading materials.
Furthermore, we have restricted the elements to structural ones

ELEMENTS INFLUENCING READING DIFFICULTY 95

which the author uses to express his ideas. This approach to the
problem in no way denies that other elements may contribute
to difficulty. Concreteness or abstractness of idea, universality
of appeal, directness of presentation, and other factors inherent
in the content unquestionably are related to ease or difficulty
of understanding. Similarly, reading interests, attitudes and
motives, intellectual capacity, previous experience, and other
factors relating to the reader probably exert large influence on
ease of reading. Important as these factors are, they are con-
sidered only incidentally in this study. We are concerned pri-
marily with reading materials and the structural elements in-
herent in their expression that are indexes of difficulty.

The investigation is limited further to materials sought by
the general reader without relation to his professional or busi-
ness interests. Fiction and general non-fiction have been con-
sidered universally applicable within certain limitations, such
as linguistic area, individual interests, and reading ability.
Since detailed studies of the reading activities of adults, as re-
ported by Gray and Monroe, show that about 50 per cent of
adults read books, 75 per cent read magazines, and 95 per cent
or more read newspapers, these three types of reading matter
have been taken as generally representative of what adults
read. 1

From the facts presented in chapter iii, it is clear that all
persons do not read equally well. Nor does any one person, as a
rule, read all materials equally well. It follows, therefore, that
not only do adults differ in ability to read, as has been em-
phasized earlier, but materials differ in difficulty for the same
reader or group of readers. To obtain an index of difficulty for
different reading materials was the first step in determining
elements of difficulty inherent in them. The second was to make
a quantitative study of the elements of expression contained in
the materials which might be related to difficulty. In the final
step, correlations were calculated between the series of elements

1 William S. Gray and Ruth Monroe, The 'Reading Interests and Habits of Adults
(New York: Macmillan Co., 1929), p. 262.

96 WHAT MAKES A BOOK READABLE

and the criterion of difficulty for each of the materials studied.
Evidence thus obtained was used to identify elements of diffi-
culty in general reading materials.

OBTAINING A CRITERION OF DIFFICULTY

In discovering how difficult general reading materials are for
adult readers, we gave the Adult Reading Test to 756 adults, as
described in chapter iii. It may be recalled that Form i of this
test contains paragraphs of fiction and Form 2, paragraphs of
non-fiction. All paragraphs in both tests were taken from the
general books, magazines, and newspapers most widely read by
adults. The method of scoring described in Appendix C renders
a score for each test paragraph which ranges from +4, the
highest possible score, to 4, the lowest possible score. Should
any test paragraph be so easy as to be perfectly comprehended
by all subjects, its criterion of difficulty would represent the
average of all perfect scores, or +4- On the other hand, if a
paragraph is so difficult that all subjects make the worst possible
responses, as they theoretically may do, the criterion of diffi-
culty, representing the average reading score, would be 4,
Average scores for all test paragraphs in Forms i and 2 of the
Adult Reading Test are given in Appendix D, on pages 334-36.
In reading these data one should remember that a high average
score designates an easy selection, and a low average score a
difficult selection,

WHAT ELEMENTS IN THE TEST PARAGRAPHS
MAY INFLUENCE DIFFICULTY?

With this question we turn again to the thesis which prompt-
ed the present study, namely, that there are elements inherent
in reading material which are related to difficulty. The average
reading scores obtained on the test paragraphs seem to support
this thesis. For example, one paragraph received an average
score of 3.0, whereas another had an average score of 0.7. Ac-
cording to the criterion of difficulty accepted in this study, the
second paragraph is clearly more difficult than the first. Accord-

ELEMENTS INFLUENCING READING DIFFICULTY 97

ing to our primary assumption, it contains elements of expres-
sion which differ in kind and quantity from those contained in
the first.

Before such elements could be determined, it was necessary
to analyze the paragraphs for the occurrence of elements of ex-
pression that may bear some relation to difficulty. In this con-
nection we attempted to compile a reasonably complete list
of the elements used in adult fiction and general non-fiction.
With this list to guide the analysis of reading materials, it was
hoped that the ultimate identification of structural elements of
difficulty would be both reliable and comprehensive.

Suggestions were derived from a variety of sources. Some
were obtained from studies of children's literature and of certain
types of adult reading. Others were suggested by directors and
teachers of adult schools who are faced with the problem of
choosing appropriate reading materials; by adult students at-
tending evening classes; and by professors of English who are
accustomed to analyze style of written discourse. These were
supplemented by a careful scrutiny of many different types of
reading matter.

Table XXVI lists the expressional elements suggested as po-
tential elements of difficulty. It may be noted that elements
numbered from i to 41 are primarily properties of words. Those
numbered from 42 to 66 belong essentially to sentences. The
remaining elements are associated generally with paragraphs of
entire selections.

EXAMINATION OF THE LIST OF EXPRESSIONAL ELEMENTS

Before accepting the elements listed in Table XXVI for use
in experimentation, it seemed essential to examine each element
critically in order to determine whether it could be adapted to
experimental procedures. What does the element mean? Is it
objectively measurable? These are the questions which we
sought to answer in examining different elements in the list.

In some cases, the first question could be answered precisely.
For example, the meaning of monosyllables, personal pronouns,

9 8

WHAT MAKES A BOOK READABLE

TABLE XXVI

ELEMENTS OF EXPRESSION SUGGESTED AS POTENTIAL INDICATORS
OF DIFFICULTY IN ADULT READING MATERIAL

NUMBER

ELEMENT

Properties of Words

*i Number of easy words

*2 Percentage of easy words

*3 Number of different hard words

*4 Number of words not known to 90 per cent of sixth-grade

children

*5 Percentage of words not known to 90 per cent of sixth-grade

children

*6 Number of different words

*7 Percentage of different words

*8 Percentage of monosyllables

*9 Percentage of bisyllables

*io Percentage of polysyllables

*n First-person pronouns

*ia Second-person pronouns

*I3 Third-person pronouns

*i4 First-, second-, and third-person pronouns

15 Nouns (proper, abstract, and concrete)

16 Verbs

17 Adjectives (descriptive, limiting)

1 8 Articles (definite, indefinite)

19 Adverbs (time, place, manner)

20 Conjunctions (co-ordinate, subordinate)

21 Interjections

22 Prepositions

*23 Content words

*24 Structural words

25 Image-bearing words

26 Non-image-bearing words

*27 Words beginning with w

*a8 Words beginning with h

*29 Words beginning with b

*3O Words beginning with i

*3i Words beginning with e

32 Words associated with adult living

33 Words expressing abstractions

*34 Asides, appositives, parenthetical expressions

*3fj Dialect words

*3 Archaic words, rare words

*37 Local expressions and coined words

38 Poetic and highly literary words

*39 Idiomatic expressions

*40 Proper nouns of mythology and history

41 Technical words

* Retained as a potential element of difficulty for further study.

ELEMENTS INFLUENCING READING DIFFICULTY 99

TABLE XXVI Continued

NUMBER

ELEMENT

Properties of Sentences

*42 Average sentence-length in words

*43 Average sentence-length in syllables

*44 Maximum syllabic sentence-length

*45 Minimum syllabic sentence-length

*4<5 Range of syllabic sentence-length

*47 Number of simple sentences

*48 Percentage of simple sentences

*49 Number of compound sentences

*o Percentage of compound sentences

*5i Number of complex sentences

*52 Percentage of complex sentences

*53 Number of compound-complex sentences

*54 Percentage of compound-complex sentences

*55 Number of compound and compound-complex sentences

*56 Clauses introduced by subordinate conjunctions

*57 Clauses introduced by conjunctive adverbs

^58 Clauses introduced by relative pronouns

*59 Prepositional phrases

*6o Infinitive phrases

*6i Prepositional and infinitive phrases

*62 Explicit thought-statements

*6j Implicit thought-statements

*64 Static predication of thought-statements

*65 Dynamic predication of thought-statements

*66 Figures of speech

Properties of Paragraphs or Entire Selections

*6j Number of words in a selection

*68 Number of sentences in a selection

*69 Number of ideas in a selection

*yo Direct discourse

*yi Indirect discourse

72 Scenic narration

73 Dramatic narration

74 Physical associations

75 Psychic associations

*j6 A sequence of ideas to show effects

*77 A sequence of ideas to show causes

*7$ Deductive presentation of a thought

*79 Inductive presentation of a thought

*8o Enumeration of ideas to show cause and effect

*8t Repetition of an idea for emphasis

*82 Development of an idea from a topic sentence

ioo WHAT MAKES A BOOK READABLE

and simple sentences is clearly fixed. In other cases, the mean-
ing is variable unless defined arbitrarily for purposes of the
study* For this reason we set up certain definitions for such
elements as "easy" words, "hard" words, "scenic" narration,
"dramatic" narration, and so on. These definitions will be given
in later sections of the chapter. In a few cases an element seemed
to defy precise definition. For example, when we attempted to
define words associated with adult living, we met the difficulty
of defining all the objectives, activities, attitudes, and other
aspects of adult living. Such an undertaking was obviously im-
possible. Since words associated with adult living, then, cannot
be defined accurately, that element was rejected from the list of
elements retained for experimentation.

Answering the first question tended generally to answer the
second. That is to say, if an element can be defined precisely,
it can be measured objectively in the majority of cases. For
example, once we have defined "easy" words, as is done farther
on, we can count them as accurately as monosyllables, the
meaning of which is established universally. If doubt arose in
determining whether an element is objectively measurable, we
made repeated counts of the element in the same sample of
reading material and compared results of the several counts. If
the counts disagreed, then it was obvious that the element does
not meet the test of objectivity.

The following paragraphs indicate the nature of the examina-
tion made of each element in Table XXVI. Wherever it seems
necessary the definition of the element is included.

Easy and hard words. In accordance with the plan adopted
by Dale and Tyler in their study of technical reading materials,
a word was here assumed to be "easy," regardless of its length,
if it is a familiar word, and "hard," if unfamiliar, 2 Familiarity
was established by the presence of a word in a composite list 3

9 Edgar Dale and Ralph W. Tyler, "A Study of the Factors Influencing the Difficulty
of Reading Materials for Adults of Limited Reading Ability," Library Quarterly, IV
tT^ly, 1934), 384-412.

3 Edgar Dale* "A Comparison of Two Word-Lists,'* Educational Research Bulletin, X
(December 9, 1931), 484-89,

ELEMENTS INFLUENCING READING DIFFICULTY 101

of the most frequent words in Thorndike's The Teachers' Word
Book 4 and the Word List of the International Kindergarten
Union. 5

Since the occurrence of a word in the Thorndike list indicates
that it is among the thousand words most frequently encoun-
tered in general reading, and since its presence in the kinder-
garten list is evidence of its familiarity to children before they
enter first grade, the conclusion seems warranted that the word
possesses zero difficulty for adults. Seven hundred and sixty-
nine words were found by Dale to be common to both lists.
These constitute the list of easy words which Dale and Tyler
have found to be indicators of ease in technical material.

Words not designated "easy" were here considered "hard/*
as in Dale and Tyler's study. Again the classification is arbi-
trary. Although it may seem unduly broad, such a classifica-
tion recognizes that an extremely common word, like "as/* "so/*
"does/' and "get," may bear connotations that make it far
more difficult than a less common word, like "bonfire" or
"truck," the meaning of which is relatively fixed. Hard words
can be accurately determined after easy words have been
checked from a given selection. The extent of hardness implied
in any word so classified was determined by the next element.

Words not known to 90 per cent of sixth-grade pupils. The in-
clusion of this element as a probable index of difficulty was sug-
gested by Dale's study of the familiarity of 8,000 common words
to elementary-school pupils. 6 Each word was classified by Dale
according to the percentage of familiarity it was found to have
among 7,878 children in the fourth, sixth, and eighth grades.
By administering the list to adults he found that their scores

4 Edward L. Thorndike, A Teachers* Word Book of the Twenty Thousand Words
Found Most Frequently and Widely in General Reading for Children and Young People.
New York: Teachers College, Columbia University, 1932. Pp. 182..

5 International Kindergarten Union Child Study Committee, A Study of the Vocabu-
lary of Children before Entering First Grade. Washington: International Kindergarten
Union, 1928. Pp. 36.

6 Edgar Dale, "Familiarity of 8,000 Common Words to Pupils in the Fourth, Sixth,
and Eighth Grades." Unpublished study, Bureau of Educational Research, Ohio State
University, Columbus, Ohio. Pp. 84.

loa WHAT MAKES A BOOK READABLE

correlated .65 with sixth-grade scores. It seems likely, there-
fore, that a word not known to 90 per cent of sixth-grade pupils
will present definite difficulty to adults of limited reading
ability. Support for this hypothesis has been presented by Dale
and Tyler in their study of elements of difficulty in technical
materials, and by Ojemann in his investigation of factors asso-
ciated with difficulty of materials used for parent education. 7

Different words. Claims have been made that range of vo~
cabulary, or vocabulary diversity, as measured by the percent-
age of different words in a selection is an indicator of difficulty.
Two assumptions are made in accepting this element as a meas-
ure of vocabulary difficulty. One is that a smaller percentage of
different words must have a greater amount of repetition. The
other is that repetition tends to make reading material less diffi-
cult. The findings of Lively and Pressey are distinctly sugges-
tive of comparative difficulty among vocabularies found in
textbooks when range of vocabulary is obtained for thousand-
word samplings. 8 By using a combined measure of frequency
and range, other investigators have obtained similar findings:
Dolch, in his study of school readers; and Patty and Painter,
in their evaluation of high-school texts. 9 Vogel and Washburne
found that the number of different words in a thousand is the
best single indicator of difficulty of a selection for school chil-
dren 3 since it correlates highest with their median reading
score. 10

* Dale and Tyler, op. cit.; Ralph H. Ojemann, "The Reading Ability of Parents and
Factors Associated with Reading Difficulty of Parent-Education Materials," Researches
in Parent Education, II. University of Iowa Studies, Studies in Child Welfare, VIII
(March I, 1934), 11-32.

8 Bertha A. Lively and S. L. Pressey, "A Method for Measuring the 'Vocabulary
Burden* of Textbooks, Educational Administration and Supervision, IX (October,
i9*3)> 389-98.

9 Edward William Dolch, "Vocabulary Burden," Journal of Educational Research,
XVII (March, 1928), 170-83; W. W. Patty and W. I. Painter, "Improving Our Method
of Selecting High-School Textbooks/' ibid., XXIV (June, 1931), 23-32; W.W.Patty
and W. I. Painter, "Technique for Measuring the Vocabulary Burden of Textbooks,"
ibid., XXIV (September, 1931), 127-34.

* Mabel Vogel and Carleton Washburne, "An Objective Method of Determining
Grade Placement of Children's Reading Material,* 7 Elementary School Journal, XXVIII
(January, 1928), 373-81.

ELEMENTS INFLUENCING READING DIFFICULTY 103

Syllabic length of word. Evidence obtained by Bear in an
experimental study of the comparative difficulty of monosyl-
lables and polysyllables in children's reading led to the conclu-
sion that long words are generally more difficult than short
words. 11

McCluskey's quantitative analysis of adult reading materials
representing six fields of subject matter showed that easy read-
ing matter, embodied in fiction, contains more short words and
fewer long words than are found in more difficult selections. 12
In content bearing on psychology and physics, which according
to objective tests proved to be most difficult, fewer short words
and more long words were discovered than in fiction. The same
elements which differentiate easy from difficult passages were
found to differentiate also moderately difficult passages from
those of still greater difficulty. The distinction was merely a
matter of degree.

Personal pronouns. The hypothesis that the informality
with which material is written increases its simplicity and con-
sequently the ease with which it is read, has led other investi-
gators to determine the relationship between number of per-
sonal pronouns and reading difficulty. That second-person pro-
nouns are indicators of ease in technical material has been found
by Dale and Tyler. For the present study, personal pronouns
in the first, second, and third person were taken first as separate
elements, and were then combined into a single element for ex-
perimentation.

Other parts of speech. The classification of words according
to their function in language expression represents the so-called
"parts of speech." They include nouns, pronouns, verbs, ad-
jectives, adverbs, conjunctions, interjections, and prepositions.
Other investigators have found significant relationship to diffi-
culty only in the case of pronouns and prepositions. In this

11 The major findings of Bear's study are summarized by George R. Johnson, "An
Objective Method of Determining Reading Difficulty," Journal of Educational Research,
XXI (April, 1930), 283-87.

12 Howard Y. McCluskey, *'A Quantitative Analysis of the Difficulty of Reading Ma-
terials." Unpublished study, University of Michigan, 1933.

io 4 WHAT MAKES A BOOK READABLE

study, therefore, these two classes of words were isolated for
enumeration. Classifications of words according to meaning and
association were substituted for classification according to func-
tion. They are described in paragraphs that follow.

Content and structural words. The classification of words as
"content" or "structural" grew out of the hypothesis that con-
tent words, namely, nouns, verbs (except copulative and auxil-
iary), descriptive adjectives, and descriptive adverbs, tend to
influence difficulty by their intrinsic power of conveying ideas.
On the other hand, structural words, namely, pronouns, copu-
lative and auxiliary verbs, limiting adjectives, articles, limiting
and conjunctive adverbs, prepositions, conjunctions, and inter-
jections were believed to bear a negative relationship to diffi-
culty since their scope of meaning is more nearly constant than
is generally true of content words.

Image-bearing and non-image-bearing words. The proposed
subdivision of content words into those that produce images
and those that do not was based on the assumption that image-
bearing words increase comprehension through their expression
of experience in the form of mental pictures. Although the enu-
meration of image-bearing words is theoretically a simple mat-
ter of cataloguing words that revive the memory of some sense
impression, it is actually a far more complex undertaking.
Clearly, the classification of a word as "image-bearing" or "non-
image-bearing" is a matter of individual experience and there-
fore subjective and unreliable. Furthermore, it seems evident
that an image-bearing word not only may arouse a simple image
but may combine with other words, each bringing its own train
of associations to create a complex image. This circumstance
obviously adds to the possibility of unreliable enumeration.

Unless individual reactions can be obtained from each reader
as to words which produce images and the number of images
produced, any classification of words as image-bearing or non-
image-bearing is quite certain to be inaccurate and unreliable.

Words beginning with "<' "A," ' V "'V or "*," That
words whose initial letter is ;, h, b> i y or e bear some relationship

ELEMENTS INFLUENCING READING DIFFICULTY 105

to difficulty is indicated by the findings of Lewerenz and of
Dale and Tyler. 13 The former found that these words vary in
number according to the ease or hardness of the material in
which they occur. That is, words beginning with w, h, or b
occur with comparative frequency in easy materials, whereas
words beginning with i or e are relatively few. In difficult
reading material the situation is reversed. Dale and Tyler found
a significant relationship between the presence of words begin-
ning with i in a given passage and its reading difficulty. Slight
relationship was noted between words beginning with other
"critical" letters and difficulty of technical passages in which
the words occurred.

Words associated with adult living. The opinion is sometimes
offered that many so-called "difficult" words are quite familiar
to the adult reader because they deal with the common affairs
of adult life. Being familiar, they are not difficult. 14 The as-
sumption seems tenable that contact with business and social
life, as adults know it, results in an understanding of words the
meaning of which is derived from such contact. The chief limi-
tation in the use of this element lies in the fact that adult ex-
periences cannot be defined in general terms to fit the unparticu-
larized adult. Before this element can be investigated scientifi-
cally, it is necessary to determine what are the common experi-
ences of groups of adults classified according to age, sex, type of
environment, amount of schooling, occupation, and so on. The
next step is to ascertain how words associated with these ex-
periences influence difficulty for various classes of readers.

Words expressing abstractions, The assumption that abstract
words complicate meaning grew out of certain observations rela-
tive to differences between the language of very simple and very
cultured people. The language of the former is characterized by
a simple concrete vocabulary acquired naturally in direct con-

Alfred S. Lewerenz, "Measurement of the Difficulty of Reading Materials," Los
Angeles Educational Research Bulletin, VIII (March, 1929), 11-16; Dale and Tyler,
ojp. cit.

** Edward William Dolch, Reading and Word Meanings (Boston: Ginn & Co., 1927),
pp. 50-51.

106 WHAT MAKES A BOOK READABLE

tact with experience. The language of cultured people, on the
other hand; contains words that represent generalizations and
abstractions. Meanings of these words result from varied group-
ings of experiences whose very difference enables their common
elements to survive in isolation. 15 In repeating several trial
counts of abstract words in a selection, we noted frequent in-
accuracies, probably for the reason that as a person becomes
familiar with the many connotations of an abstract word the
quality of abstractness is gradually lost. It seems clear that
unless a word can be defined precisely as an abstraction, it can-
not be considered objectively measurable.

Asides, appositives, and parenthetical expressions. Since ex-
tremely elementary material tends to utilize words, phrases, and
clauses of co-ordinate rank, it was believed that subordination
of certain elements of a thought-unit would present difficulties
for the reader of immature reading habits.

Other expressional elements related to words. Other elements,
numbered from 35 to 41 , inclusive, in Table XXVI were be-
lieved to influence difficulty, on the assumption that words
which have acquired meaning for persons in one environment
may have relatively little significance for others. It seems prob-
able that four of these elements can be selected from reading
material with considerable accuracy. They are "dialect" words,
which are sometimes italicized; "archaic" expressions and
"rare" words, which can be recognized by their unusualness;
"local" expressions and "coined" words, frequently indicated by
quotation marks; and proper nouns pertaining to mythology
and history.

Other elements appear to be less easily recognized. For ex-
ample, to designate a word as "highly literary" involves subjec-
tive judgment and tends to result in unreliable enumeration.
This element was rejected, therefore, despite the fact that Ayer
has shown in her study of difficulties in reading history that

*s C. K. Ogden and I. A, Richards, The Meaning of Meaning (New York: Harcourt,
Brace & Co., 1927), p. 213.

ELEMENTS INFLUENCING READING DIFFICULTY 107

literary embellishments baffle comprehension for children. 16 An
accurate enumeration of idiomatic expressions seems impos-
sible, for the reason that they have become characteristic of our
language and are not readily identified. They were rejected to-
gether with technical words which presumably occur infrequent-
ly in general, non-technical material.

Length and kind of sentences. Studies of elementary-school
readers indicate that authors of these books attempt to adapt
materials to different grade levels by varying the length and
complexity of sentences. For example, Harris found a constant
increase in average length of sentences from the first to the sixth
grades, although the amount of increase varied at different
levels. 17 She also discovered a decrease in the number of simple
sentences and an increase in the number of complex sentences
over the same school period. A marked increase was further
observed in the occurrence of compound and compound-com-
plex sentences, although the percentage was smaller.

Investigators who have attempted to determine the relation-
ship between length and kind of sentences and difficulty of read-
ing materials have reported findings that are not altogether
consistent. In the field of adult reading, Dale and Tyler found
a significant relationship between length of sentence in words
and difficulty of comprehension of technical materials. The re-
lationship was noticeable, though not statistically significant,
between types of sentences and reading difficulty. McCluskey
discovered relatively short sentences in books of easy fiction,
but longer and more complex sentences in selections from psy-
chology and physics.

In the field of children's reading, Vogel and Washburne found
simple sentences one of the most reliable indexes of easy reading
material. Two other investigators, Orndorff and Thompson,
agree that although a short sentence may require less effort to

rf Adelaide M. Ayer, Some Difficulties in Elementary School History (New York: Bu-
reau of Publications, Teachers College, Columbia University, 1926), p. 10.

J 7 Helen C. Harris, "The Development of Language Ability during the Elementary
School Period." Unpublished Master's thesis, Dept. of Education, University of Chi-
cago* I 93 < 5- Pp- 105-

io8 WHAT MAKES A BOOK READABLE

read, sentence-length is of no consequence so far as comprehen-
sion is concerned. 18 Their findings indicate that the effect upon
the reader is the same, whether the sentence is long and involved
or short, simple, and emphatic. Several weaknesses noted in
OrndorfFs study probably tend to invalidate the findings. The
number of cases is small. Speed is emphasized. The test ma-
terials are unstandardized and too long for the attention-span
of sixth-grade children. Using the same test twice in succession
appears to be a further weakness.

In Thompson's study, a probable weakness lies in the use of
narrative material only. An examination of the selections used
seems to show an overpotency of interest and excitement, which
may so illuminate certain high spots in the reading as to give
cues for desired responses to the tests. Had a second paragraph
been added, the comprehension of which depended on the pu-
pil's ability "to select, weigh, compare, and organize the ele-
ments of the paragraph," 19 the results might have shown closer
agreement with the more extensive studies cited earlier.

Five measures of sentence-length were used in the present
study. All of them may be measured objectively. The number
of words in a sentence was the measure of length usually
considered. Length expressed in number of syllables was also
used on the assumption that a measure by component parts is
more discriminative than a measure by the sum of an indefinite
number of parts. Other measures of sentence-length that are
variations of the same element include "maximum syllabic"
length, that is, the length of the longest sentence; "minimum
syllabic" length, that is, the length of the shortest sentence; and
"range of syllabic" length, which refers to the number of syl-
lables between the longest and the shortest sentence in a selec-
tion.

18 Bernice OrndorfF, "An Experiment to Show the Effect of Sentence Length upon
Comprehension." Master's thesis, Dept. of Education, University of Iowa, 1925. Pp.
80; Ruth Culver Thompson, "The Effect of Length of Sentence upon Comprehension/*
Master's thesis, University of Pittsburgh, 1929. Pp. 40.

x Edward L. Thorndike, "Reading as Reasoning," Journal of Educational PsychoL
ogy> VIII (June, 1917), 329.

ELEMENTS INFLUENCING READING DIFFICULTY 109

Types of subordinate clauses. Variants in expression which
represent types of subordinate clauses were suggested as prob-
able indicators of difficulty. Subordinate clauses introduced by
subordinate conjunctions that, so that, if, in order that and
clauses introduced by conjunctive adverbs when, where, how
involve a suspension of judgment as to the outcome of the
sentence. It was believed that they would cause greater diffi-
culty than subordinate clauses introduced by relative pronouns
who, which,, that.

Kinds of phrases. In the small and concrete vocabulary of
very simple people, few phrases have been noted. This observa-
tion led to the assumption that phrases complicate content and
are, therefore, elements of difficulty. In their study of technical
material, Dale and Tyler found that the presence of preposi-
tional phrases increased its difficulty significantly, either by
their effect on complexity of thought or on sentence-length.
Vogel and Washburne found that the number of prepositions in
a selection was one of the four best indicators of difficulty in
children's books.

The question of predication. Grammarians used to say that
predication was the one essential condition in sentence-forma-
tion. Either predication was expressed or it was implied in the
context. The modern point of view is quite different. Predica-
tion is no longer considered necessary. If it is actually expressed,
we say that the sentence is "explicit/' If, on the other hand,
elements of the sentence are omitted subject, predicate, modi-
fiers and the thought-relationships only implied, the sentence
is "implicit. 1 ' Disregard for predication is illustrated by the
following brief passage, in which three out of five sentences
leave elements to be supplied by the imagination of the reader:

He wore overalls, sweater, an old felt hat. A man of the soil. Obviously.
But his eyes; they were of the sky! Blue, searching, steady eyes. 20

By analyzing selections for predication, we hoped to discover
whether the expression or implication of thought-relationships
influence difficulty in the same degree and in the same direction.

80 Item No. ii, Form i, of the Adult Reading Test.

no WHAT MAKES A BOOK READABLE

Two other variants of expression were suggested by the ques-
tion of predication. They concern different modes of predica-
tion, namely, "dynamic" predication, which implies a change of
condition; and "static" predication, which merely associates
ideas without implying a change of condition. The one expresses
a thought by the use of verbs of action in the active voice, while
the other utilizes passive verbs and verbs expressing state or
continuance. These two modes of expression are commonly con-
sidered stylistic devices introduced for desired effects. It was
believed, however, that one or the other might bear significant
relationship to difficulty of comprehension. This hypothesis,
like many others, had its origin in a study of the evolution
of human speech. Research in this field shows that in prim-
itive speech the dominant mode of predication employs action
words, which bear the character of human mood and human
will. 21 It may be assumed, therefore, that dynamic predica-
tion promotes understanding more than does static predica-
tion.

What do the author V words mean? This question arises when
understanding is based not alone on primary denotations of
words but on associations introduced for the purpose of clarity,
intensity, or enrichment. These associations are commonly re-
ferred to as "figures of speech/* Whereas deficiencies in literal
meaning may be remedied by the use of the dictionary or en-
cyclopedia, obstacles to the understanding of figurative mean-
ings associated with the same words are less readily remedied.
On the assumption that figurative language influences the diffi-
culty of reading, five figures of speech most common in general
material were suggested as probable indicators of difficulty.
They are simile, metaphor, metonymy, synecdoche, and per-
sonification.

Length of selection. There is some evidence that brevity of
statement is essential to comprehension for the inexperienced

21 Bronislaw Malmowski, 'The Problem of Meaning in Primitive Languages," cited
in Ogden and Richards, The Meaning of Meaning (New York: Harcourt, Brace & Co.,
> PP-

ELEMENTS INFLUENCING READING DIFFICULTY in

reader. There is also evidence that it is the better reader who
chooses large books, long chapters, or detailed treatises. Dale
and Tyler report some relationship between the difficulty of
technical material and the number of words contained therein.
Although McCluskey reports a similar relationship, he con-
cludes that the number of ideas in a selection is of less conse-
quence in determining difficulty than is the quality of the ideas.
The reliability of his conclusion depends for the most part on
the accuracy with which the ideas were counted, since ideas
probably cannot be counted with the same degree of objectivity
as can words and sentences.

Direct and indirect discourse. The assumption is generally
made that quoting the words of another, either directly or in-
directly, tends to make a selection more easily comprehended.
It seems probable that direct discourse is more closely related
to ease than is indirect discourse. The greater understanding is
explained by the fact that in reading the former, a reader tem-
porarily projects himself into another personality and expresses
his sentiments and ideas. Hence he may gain greater under-
standing of their meanings.

Types of narration, 22 Literary criticism, when directed
toward the structure of a novel or a story, recognizes various
techniques utilized by the author to produce desired effects.
They may be techniques of narration "scenic," "pictorial,"
"panoramic," and "dramatic." The first three terms seem to be
used interchangeably to describe a single type of narration.
They refer to the author's attempt to portray a scene without
action, merely by reflecting events and pictures upon his read-
er's consciousness. Let him, however, take from the scene its
suggestion of dialogue, its people, their dress, their actions all
the actual things which he has described and place them on a
stage, so to speak, and he produces a new effect. He has intro-
duced dramatic narration.

32 Percy Lubbock, The Craft of Fiction, New York: Jonathan Cape and Harrison
Smith, the Traveller's Library, 1929. Pp. 277; Ralph Philip Boas, The Study and Ap-
preciation of Literature (New York: Harcourt, Brace & Co., 1931), p. 152.

ii2 WHAT MAKES A BOOK READABLE

For the most part, narration is a question of the reader's rela-
tion to the author. By the scenic method the reader is placed
face to face with the story-teller, to listen to him report a scene.
By the other, the reader is turned toward the story to watch its
action. The hypothesis seems warranted that the conversation-
al, active, dramatic method of narration should be less difficult
to comprehend than the method which leaves the passive reader
to make what he can of a narrated event.

Literary criticism, however, is seldom mathematically exact.
Hence it is difficult to define the two methods so sharply that
inaccuracies of classification will be obviated.

Presentation of character. Two stylistic devices which were
suggested as indexes of difficulty relate to a story-teller's meth-
od of presenting character. The first involves the use of physi-
cal associations. This method is found in the older type of novel.
It depicts a character by the use of sense impressions and ex-
periences drawn from the physical environment. It is believed
to promote better understanding than the modern method of
developing a character by the presentation of his psychic life
his thoughts, impulses, and emotions. There is a straightfor-
wardness about the first method that gives tangibility to a
character, while in the second there is an evasiveness that
admits of little substance. It seems probable, therefore, that
readers of low ability may experience more difficulty in under-
standing portrayal of character by the latter method than by
the former.

Techniques of paragraph-development. The last seven ele-
ments listed in Table XXVI relate to an author's technique of
developing paragraphs. He may present a sequence of ideas to
show cause or effect. He may develop an idea deductively or
inductively. He may enumerate ideas to show both cause and
effect, or repeat an idea for emphasis, or develop it from a topic
sentence. In considering each of these techniques we have as-
sumed that some of them promote comprehension and others
retard it.

ELEMENTS INFLUENCING READING DIFFICULTY 113

SELECTION OF ELEMENTS TO RETAIN FOR
EXPERIMENTATION

The statement has been made earlier that lack of evidence
relating to elements of difficulty in general reading materials
demands that the person who is interested in this field of investi-
gation remain open-minded toward expressional elements that
suggest some influence on difficulty. Consequently, the selec-
tion of elements listed in Table XXVI for further study de-
pended primarily on preciseness of definition and objectivity
of measurement.

Elements which seem to defy objective measurement in-
clude: image-bearing and non-image-bearing words, words as-
sociated with adult living, abstractions, poetic and highly liter-
ary words, scenic and dramatic narration, and physical and
psychic associations. Consequently, they were rejected from
the present study. Technical words were classified in the gen-
eral category of hard words. Elements representing the classifi-
cation of words according to function, with the exception of per-
sonal pronouns and prepositions, were rejected in favor of ele-
ments representing classifications of words according to mean-
ing and association. The remaining sixty-four elements, which
survived critical examination, were retained for further experi-
mentation. These elements are indicated by asterisks in Table
XXVI.

Use of the compiled elements. Although considerable evidence
has been reported to indicate that certain of the elements bear
some relation to difficulty, it has been obtained principally in
the field of children's reading or in the field of technical reading
on the adult level. Evidence is contradictory or wholly lacking
concerning other elements. In any case, the influence of a spe-
cific element on difficulty cannot merely be assumed to exist for
general reading material. Accordingly, each element in the table
was considered no more than a potential element of difficulty
until its relationship to a criterion of difficulty could be calcu-
lated. As previously stated, the criterion accepted in the present

1 1 4 WHAT MAKES A BOOK READABLE

study was the average reading score of adult readers on Forms i
and a of the Adult Reading Test.

Determining the occurrence of each element in the various
test items involved a simple count of each element. This was
expressed as a numerical total or as a percentage. Of the sixty-
four elements used in the study, forty-four occurred one or
more times in at least half of the test items. A less frequent oc-
currence was considered inadequate for correlation. The rejec-
tion of the other twenty elements leaves their influence on diffi-
culty an open question. Whether their occurrence in an occa-
sional test item tended to increase the difficulty of that item
remains to be determined. Limitation of time and effort would
not permit extension of the study to include a larger number of
items containing these twenty elements.

CORRELATING THE CRITERION OF DIFFICULTY WITH POTEN-
TIAL ELEMENTS OF DIFFICULTY IN THE TEST ITEMS

In accordance with the plan outlined earlier, a series of cor-
relations was calculated to show the relationship between the
elements in the test items and the average reading score of 756
adult readers. Since the primary purpose of the study was to de-
termine elements of difficulty in general reading material, the
forty-eight items in the two tests were combined into a single
series for purposes of correlation.

The first column in Table XXVII presents the coefficients of
correlation obtained when the average reading score of all adults
was taken as the criterion of difficulty. If the 756 adults tested
are fairly representative of the general reading public, then ele-
ments identified by using their average score may be regarded
as elements of difficulty for the general reader. The second col-
umn in the table shows the relationship between the elements in
the test items and the average reading scores of the "best"
readers. They were the readers whose combined scores on Forms
I and a of the adult test ranked above the third quartile. Their
average score was a criterion of the difficulty which they ex-
perienced in reading the test paragraphs. In the third column

ELEMENTS INFLUENCING READING DIFFICULTY 115

TABLE XXVII

CORRELATIONS BETWEEN FORTY-FOUR POTENTIAL ELEMENTS OF DIFFICULTY

AND THE AVERAGE SCORE OF EACH TEST ITEM FOR THREE

CLASSIFICATIONS OF ADULT SUBJECTS

NUM-
BER

ELEMENT

CORRELATION COEFFICIENTS

All Readers

"Best"
readers

"Poorest"
Readers

r P.E.

i
i

4
6

7
8

12
13

17
18

*3
2 4

Average sentence-length in words
Percentage of easy words

.522i .0708

.520+ .0710

-.5I 3 . 0717
.511+. 0719
-.496 -0734

--49I -0739
.482 .0747

.4761.0753
-.4741.0755

-.472+ .0757
433^79 I

.431 -0793

.402! .0816
.390 .0825
-.3801.0833
-.380+ .0833
-.345- 8 5 8

.329! ,0868

-.321 .0873
.291 .0891

.283 .0896
.261 .0907
.261 .0907
.235 .0920

--43 1 1-0793
.27i .0902

.21 8 .0927
.292+ .0891
.1911.0938

-.481 .0748

.380+ .0833
.3471.0856
-.3021.0885

.289! .0892
.1661.0947

.3371.0863

-.2271.0923
.389! .0826
.1811.0942
-.0771.0968
.0621.0970

.1501.0952

.1141.0961
-.0751.0968

.0501.0971
-.1731.0944

173 1.0944
.1201.0960

-.4491-0777
.5291.0701

-.55l- 6 79
.518! .0712
-.543i-o686

-.4061.0813
.419! .0803

.450! .0776
.4071.0812

-.457 1.0770
.4881.0742

.3711.0840

.4091.0811
.2981.0887
-.4411.0784
-.4281.0795
-.37i-o 8 4

.3071.0882

.2791.0898
-.3671.0842

.2711.0902
-.2321.0921
.2321.0921
.2851.0894

Number of words not known to
90 per cent of sixth-grade pu-
pils .,..'.

Number of easy words

Number of different hard words
Minimum syllabic sentence-
length

Number of explicit sentences. . .
Number of first-, second-, and
third-person pronouns

Maximum syllabic sentence-
length

Average sentence-length in syl-
lables

Percentage of monosyllables . . .
Number of sentences per para-

crraDh

Percentage of different words
not known to 90 per cent of
sixth-grade pupils

Number of simple sentences
Percentage of different words. .
Percentage of polysyllables
Number of prepositional phrases
Number of third-person pro-
nouns

Range of syllabic sentence
length

Number of different words
Number of infinitive and prepo-

Percentage of content words. . .
Percentage of structural words.
Number of i words

ii6 WHAT MAKES A BOOK READABLE

TABLE XXVII Continued

NUM-
BER

ELEMENT

CORRELATION COEFFICIENTS

All Readers

"Best"

Readers

"Poorest"
Readers

r P.E.

*5
26
27

28
29
3
3i

32
33
34

3 I
36

38
39

41
42

43
44

Number of figures of speech. . . .
Percentage of complex sentences
Number of compound-complex
sentences

.2331.0921

.229 .0922
.205 .0933

I99-0935
i94l.937
i93-937
~.I93 -0937
.1831.0941
.180+ .0942

,160! .0949
.1461.0953

.141 .0954

.117! .0960
-.0971.0964

.0911.0965
.069! .0969
-.0651.0969

.060! .0970
.0541.0971
-.0091.0973

.088! .0966
-.3591.0848

.2491.0913
.038! .0972
.272! .0902
.0061.0974
.1891.0939
.0511.0971
.2781.0898

-.1981.0935
.045-97 a

*94-937

.0281.0973
.0791.0967

.1681.0946
004! .0974
,1981.0935

.114! .0961
-.0181.0973
.070! .0969

-377 -0835
-.0931.0965

1381.0955
.300! .0886
.mi .0962
.255! .0910
-.1871.0939
.168! .0946
.0961.0965

~.i77-943
.1271.0958

.0661 .0969

.2341.0920
.2111.0930

.060! ,0970
.093! .0965
-.153 -0951

.0231,0973
.0191.0973
.0751.0968

Number of infinitive phrases. . .
Number of first-person pronouns
Number of complex sentences. .
Percentage of bisyllables

Number of h words

Percentage of simple sentences.
Number of clauses introduced
by subordinate conjunctions.
Number of w words . . . .

Number of compound and com-
pound-complex sentences. . . .
Number of clauses introduced
by relative pronouns

Number of asides

Total number of words per para-
graph

Number of b words

Number of e words

Percentage of compound-com-
plex sentences

Number of clauses introduced
by conjunctive adverbs

Percentage of compound and
compound-complex sentences

of the table are shown the coefficients obtained when the ele-
ments were correlated with the average reading scores of
"poorest" readers, that is, of those individuals whose combined
scores ranked below the first quartile. The average scores of this
group represented the difficulty which the paragraphs held for
its members. A comparison of the three series of correlations
indicates whether an element of expression which influences

ELEMENTS INFLUENCING READING DIFFICULTY 117

difficulty for "poorest" readers influences difficulty to the same
degree for "best" readers, when each is isolated and when all
readers are combined.

In the interpretation of the coefficients in Table XXVII, one
needs to remember that the higher the average score on an item
the greater the group comprehension. High scores indicate items
easily comprehended and low scores indicate items difficult to
comprehend. It follows, therefore, in the use of these scores for
correlation, that a negative coefficient means that the element
is positively correlated with difficulty. A positive correlation,
on the other hand, means that the element is negatively corre-
lated with difficulty. Stated in more simple terms, a negative
coefficient designates an element of difficulty; a positive coeffi-
cient 3 an element of ease.

The amount of correlation is indicated by the size of the cor-
relation coefficient. All of the correlations may seem low, as one
commonly thinks of correlation, yet their significance is to be
interpreted in the light of their relative size rather than their
absolute size. Of the correlations listed in the table, only the
first twenty-one are statistically significant, since a coefficient
must be at least .27 to be three times its probable error. Actu-
ally, however, if the occurrence of an element has even a small
numerical relationship to difficulty, nothing can seemingly justi-
fy indifference to it, since unquestionably the relationship didn't
"just happen."

WHAT ELEMENTS INFLUENCE DIFFICULTY OF GENERAL

READING MATERIALS FOR DIFFERENT

CLASSES OF READERS?

We are now ready to answer one of the questions which this
chapter aimed to answer. The final evidence for identifying ele-
ments of difficulty is found in the table just presented. In using
the relationships listed in the table for this purpose, we are as-
suming a causal relationship which by the nature of a limited
investigation may not be altogether valid. Inasmuch as all fac-
tors have not been controlled, it is possible, though hardly prob-

ii8 WHAT MAKES A BOOK READABLE

able, that the relationships may be accounted for by some other
factor operating throughout. One may argue, for example, that
personal pronouns are related to difficulty in a negative direc-
tion because they occur in content which an author presents
directly. Hence, it is the direct presentation that is the indi-
cator of ease and not the personal pronouns. The shortcomings
of a series of findings based on partially controlled procedures
have already been admitted in chapter i. It is in the light of
these shortcomings that elements of difficulty are identified.

A cursory examination of entries in Table XXVII shows that
in general the hypothesis set up at the outset of the study is
valid, namely, that elements representing simplicity of expres-
sion correlate in the direction of ease, and that those represent-
ing complex, involved expression correlate in the direction of
difficulty. Some exceptions to this observation may be noted.
For example, we had assumed that complex and compound-
complex sentences are indicators of difficulty, since they repre-
sent involved thought-patterns. Such was found to be true,
however, only when their percentage of occurrence among all
kinds of sentences was considered.

Interesting outcomes resulted from analyzing these sentences
with respect to the nature of their subordinate clauses. Then it
was found that clauses introduced by relative pronouns are
slight indicators of ease. Clauses introduced by conjunctive ad-
verbs give some indication of difficulty. Clauses introduced by
subordinate conjunctions are indexes of difficulty to a greater
degree and for all classes of readers. It appears from this evi-
dence that mere complexity of sentence-form has less influence
on difficulty than may be presumed. Moreover, the influence
is less than particular modes used in effecting subordination. It
appears also that subordinate clauses which delay comprehen-
sion by setting up restrictions of manner, degree, and condition
offer greater difficulty for all classes of readers than do other
types of subordinate clauses. They are the ones, then, that
should be used infrequently in preparing material for readers of
limited ability.

ELEMENTS INFLUENCING READING DIFFICULTY 119

When the average reading score of all adults was taken as the
criterion of difficulty, twenty-three elements were discovered
with a relationship to difficulty ranging from .522 for average
sentence-length in words to .009 for percentage of compound
and compound-complex sentences. Nineteen elements were
found to correlate positively with difficulty for "best" readers.
These are indicated in the second column of Table XXVII by
negative coefficients ranging from 481 for minimum syllabic
sentence-length to .018 for clauses introduced by conjunctive
adverbs. Twenty-four elements of difficulty were found for
"poorest" readers. The coefficients of correlation range from
.550 for number of words not known to 90 per cent of sixth-
grade pupils to .019 for clauses introduced by conjunctive
adverbs.

Most of the elements which show relatively low correlation
with difficulty occurred in no more than half the test items.
This was the minimal occurrence set up arbitrarily for correla-
tion. For example, number of asides, correlating positively with
difficulty for all but "best" readers, occurred in only twenty-
eight of the forty-eight test items. Furthermore, the occur-
rence in each item was low, ranging between i and 3. A simi-
larly restricted representation was noted for compound-complex
sentences, different kinds of subordinate clauses, figures of
speech, infinitive phrases, and first-person pronouns.

If we speculate on how much significance should be attached
to a low correlation when only a few elements are involved, it
seems evident that it has little or no potency beyond the sug-
gestion that a relationship exists. The assumption seems tenable
that a more significant relationship might have been discovered
had these elements occurred as frequently as did monosyllables,
for example. On the other hand, if the materials used as sources
of test items are fairly representative of what adults are read-
ing, we may infer that elements found in them tend to occur
with a frequency that is "normal" for general materials. That
is to say, although asides, for example, might have correlated
more closely with difficulty had they occurred in a larger num-

i ao WHAT MAKES A BOOK READABLE

ber of items or with greater frequency in a few items, it is pos-
sible that their occurrence represents what is usual in general
reading material. If such is the case, then the correlations given
in Table XXVII presumably signify not far from the true rela-
tionship between expressional elements and difficulty of general
materials.

One may assume, therefore, that the normal, occasional oc-
currence of some elements may contribute to difficulty, yet
their influence be too slight and their occurrence too infrequent
to command serious consideration. Further study is needed be-
fore this assumption can be verified.

A comparison of the three series of correlations in Table
XXVII shows the extent to which expressional elements agree
in their relationship to difficulty for different classes of readers.
For example, elements pertaining to sentence-structure corre-
late more closely with difficulty for "best" readers than do other
elements. For "poorest" readers, on the other hand, elements
of vocabulary correlate most closely with difficulty.

It is apparent from the size of the coefficients that the amount
of correlation tends to vary in the direction of increasing diffi-
culty for "poorest" readers, and that elements which correlate
either positively or negatively with difficulty tend to be of least
consequence for "best" readers. In other words, an expressional
element, such as percentage of different words, is an element of
difficulty for all classes, but its relation to difficulty is markedly
more significant for "poorest" readers. Again, percentage of
easy words tends to reduce the difficulty of a selection for all
classes, but its influence is greatest for "poorest" readers.

In Tables XXVIII, XXIX, and XXX are listed the forty-
four elements of difficulty in their order of significance for dif-
ferent classes of readers. As might be expected, more elements
are significantly correlated with difficulty in either a positive or
a negative direction for "poorest" than for "best" readers.
"Significant" elements, that is, those which correlate with diffi-
culty to a degree of .ij or higher, number twenty-four for
"poorest" readers and thirteen for "best" readers. Elements of

ELEMENTS INFLUENCING READING DIFFICULTY 121

little significance, that is, those whose correlation with difficulty
is below .11, number nine for "poorest" and fourteen for "best"

TABLE XXVIII
SIGNIFICANCE OF ELEMENTS OF DIFFICULTY FOR ALL READERS

Elements of Greatest
Significance

Elements of Some
Significance

Elements of Little
Significance

Average sentence-length in

Percentage of content

Number of asides

words

Total number of words per

Percentage of easy words

Percentage of structural

paragraph

Number of words not known

words

Number of b words

to 90 per cent sixth-grade

Number of i words

Number of e words

pupils

Number of figures of speech

Percentage of compound-

Number of easy words

Percentage of complex sen-

complex sentences

Minimum syllabic sentence-

tences

Number of clauses intro-

length

Number of compound-

duced by conjunctive

Number of explicit sentences

complex sentences

adverbs

Number of first-, second-,

Number of infinitive

Percentage of compound

and third-person pronouns

phrases

and compound-complex

Maximum syllabic sentence-

Number of first-person pro-

sentences

length

nouns

Average sentence-length in

Number of complex sen-

syllables

tences

Percentage of monosyllables

Percentage of bisyllables

Number of sentences per

Number of h words

paragraph

Percentage of simple sen-

Percentage of different

tences

words not known to 90

Number of clauses intro-

per cent of sixth-grade

duced by subordinate

pupils

conjunctions

Number of simple sentences

Number of w words

Percentage of different words

Number of compound and

Percentage of polysyllables

compound-complex sen-

Number of prepositional

tences

phrases

Number of clauses intro-

Number of third-person pro-

duced by relative pro-

nouns

Range of syllabic sentence-

length

Number of different words

Number of infinitive and

prepositional phrases

readers. It is clear that more elements need to be taken into
account when selecting or preparing reading matter for limited

WHAT MAKES A BOOK READABLE

readers than for superior readers. Furthermore, the elements
mean considerably more for the former group than for the latter,
as has been indicated by the coefficients in Table XXVII.

TABLE XXIX
SIGNIFICANCE OF ELEMENTS OF DIFFICULTY FOR "BEST" READERS

Elements of Greatest
Significance

Elements of Some
Significance

Elements of Little
Significance

Minimum syllabic sentence-

Number of compound-

Number of figures of

length

complex sentences

speech

Average sentence-length in

Percentage of different

Percentage of polysyl-

words

words not known to 90

lables

Number of simple sentences

per cent of sixth-grade

Number of different words

Number of explicit sentences

pupils

Number of asides

Percentage of complex sen-

Number of clauses intro-

Percentage of compound

tences

duced by subordinate

and compound-complex

Number of first-, second-.

conjunctions

sentences

and third-person pronouns

Number of e words

Number of prepositional

Number of sentences per

Number of compound and

phrases

paragraph

compound-complex sen-

Number of h words

Maximum syllabic sentence-

tences

Number of infinitive and

length

Number of different hard

prepositional phrases

Number of easy words

words

Number of w words

Average sentence-length in

Percentage of bisyllables

Number of infinitive

syllables

Percentage of different

phrases

Percentage of simple sen-

words

Number of clauses intro-

tences

Percentage of content

duced by conjunctive

Number of first-person pro-

words

adverbs

nouns

Percentage of structural

Number of complex sen-

Percentage of easy words

words

tences

Total number of words

Number of b words

per paragraph

Percentage of monosyl-

lables

Number of third-person

pronouns

Number of i words

Range of syllabic sentence-

length

Percentage of compound-

complex sentences

When the differences between the coefficients of correlation
for any two classes of readers are examined statistically, it is
found that the most significant difference occurred in figures of

ELEMENTS INFLUENCING READING DIFFICULTY 123

speech. The difference in difficulty exerted by this element for
"best" and for "poorest" readers is 3.64 times its probable

TABLE XXX
SIGNIFICANCE OP ELEMENTS OF DIFFICULTY FOR "POOREST" READERS

Elements of Greatest
Significance

Elements of Some
Significance

Elements of Little
Significance

Number of words not known to
90 per cent of sixth-grade pupils
Number of different hard words
Percentage of easy words
Number of easy words
Percentage of monosyllables
Average sentence-length in syl-
lables

Average sentence-length in words
Number of first-, second-, and

third-person pronouns
Percentage of different words
Percentage of polysyllables
Number of explicit sentences
Percentage of different words not
known to 90 per cent of sixth-
grade pupils

Maximum syllabic sentence-length
Minimum syllabic sentence-length
Number of figures of speech
Number of sentences per para-
graph

Number of prepositional phrases
Number of different words
Number of third-person pro-
nouns

Number of infinitive phrases
Number of simple sentences
Number of * words
Range of syllabic sentence-length
Number of infinitive and preposi-
tional phrases

Number of complex
sentences

Number of clauses in-
troduced by relative
pronouns

Percentage of structur-
al words

Number of asides

Percentage of bisyl-
lables

Number of clauses in-
troduced by subor-
dinate conjunctions

Number of h words

Number of e words

Number of compound-
complex sentences

Number of w words

Number of first-person
pronouns

Percentage of simple
sentences

Percentage of complex
sentences

Number of I words

Percentage of compound
and compound-com-
plex sentences

Total number of words
per paragraph

Percentage of com-
pound-complex sen-
tences

Number of clauses in-
troduced by conjunc-
tive adverbs

error. A difference of slightly less significance statistically is
noted for different hard words. This difference is 3.03 times its
probable error.

Despite the fact that the difference in relationship of many

124 WHAT MAKES A BOOK READABLE

elements to difficulty is not statistically significant for "best"
and "poorest" readers, it seems large enough for practical im-
plications. For example, percentage of easy words, number of
words not known to 90 per cent of sixth-grade pupils, percent-
age of monosyllables, percentage of polysyllables, number of
prepositional phrases, and number of/ words are related to diffi-
culty in the same direction for "best" as for "poorest" readers,
but in an amount considerably less. One may infer from a com-
parison of these relationships that reading ability ultimately
may reach a level at which comprehension is uninfluenced by
the author's mode of expression. Before that level can be iden-
tified, it is necessary to discover how well adults who are better
readers than our "best" are able to read general reading mate-
rials. If we find little or no relationship between their compre-
hension and the occurrence of structural elements in the pas-
sages read, then we may conclude that for such readers diffi-
culty cannot be measured by elements of structure. Presum-
ably other aspects should be considered in selecting and pre-
paring materials which they will find readable. For the less ma-
ture reader at lower levels of efficiency, however, the quality of
ease or difficulty apparently is of considerable importance in de-
termining whether or not a book is readable.

A COMPARISON OF THE FINDINGS WITH THOSE OF
SIMILAR INVESTIGATIONS

In comparing elements of difficulty as they have been defined
in related studies, we have included only those studies which
utilized an experimental procedure similar to the one reported
in this chapter. Three seem comparable to the present investi-
gation. The first, by Vogel and Washburne, laid the foundation
for an objective determination of elements of difficulty in chil-
dren's literature. The second, by Dale and Tyler, is a scientific
attack on the problem of difficulty of technical material. The
third, by Ojemann, is a study of the reading ability of parents
and of factors associated with reading difficulty of materials
used for parent education. All the studies identify elements of

ELEMENTS INFLUENCING READING DIFFICULTY 125

difficulty in terms of coefficients of correlation. The relative
influence of different elements rather than the actual influence
is used in comparing the present findings with those obtained
from the other three. The reason for this is that the range of
correlation varies for the different studies.

Elements of difficulty as defined by all studies generally cor-
relate more closely with difficulty for children than for adults.
The same elements correlate less closely with difficulty of tech-
nical materials for adults than with difficulty of general adult
reading matter. This latter correlation, in turn, is less than for
materials used in parent education. For example, easy words,
based on Thorndike's first 1,000, has a correlation of .674 with
difficulty of children's reading and of .640 with difficulty of par-
ent-education materials. Percentage of easy words, based on
Thorndike's first 1,000 and the Kindergarten Word-List, cor-
relates .520 with difficulty of general reading material and .352
with difficulty of technical materials.

It is important to note that the best indicator of difficulty of
general materials for "poorest" readers is the second-best indi-
cator of difficulty of technical materials and of parent-educa-
tion materials. This element, number of words not known to 90
per cent of sixth-grade pupils, correlates .550 with difficulty
of general books, magazines, and newspapers; .380 with diffi-
culty of technical reading in the field of health; and .730 with
difficulty of subject matter dealing with parent education. It
seems equally important to observe that in all types of adult
reading studied, complexity of sentence-form, as measured by
occurrence of complex sentences, bears a relatively low relation-
ship to difficulty. On the other hand, simplicity of sentence-
form, as measured by the presence of simple sentences, is a sig-
nificant indicator of ease in all adult materials.

Three elements of the fifteen which correlate most closely
with difficulty of general materials for "poorest" readers are
common to all types of material They include number of dif-
ferent hard words, number of prepositional phrases, and num-
ber of different words. Four others are common to the various

126 WHAT MAKES A BOOK READABLE

types of adult material studied. They are number of words not
known to 90 per cent of sixth-grade pupils, percentage of mono-
syllables, percentage of easy words, and average sentence-
length in words. These seven elements, then, may be thought
of as elements most closely related to difficulty of adult materi-
als generally. Of less significance are percentage of complex sen-
tences, percentage of compound-complex sentences, number of
first-person pronouns, and percentage of compound and com-
pound-complex sentences.

WHAT ELEMENTS IN GENERAL READING SHALL BE USED
IN ESTIMATING DIFFICULTY?

The practical importance of identifying elements of difficulty
lies in their usefulness in determining what materials are easy
or difficult for adults to read and understand. Inasmuch as the
results of 1 this study are intended primarily to benefit adults of
limited reading ability, it seems desirable to determine diffi-
culty of material for such readers. "Poorest" readers, that is,
adults whose combined scores on Forms I and a of the Adult
Reading Test ranked below the first quartile, have been taken
as representative of readers of limited ability. Elements of diffi-
culty for this group are shown in Table XXVII on pages 1 15-16.

Since the "poorest" readers numbered but 190, the question
may be raised as to whether this number is an adequate sample
of total population which they are taken to represent. We have
accordingly obtained a reliability coefficient by splitting the
group into random halves and correlating the average score on
each test item calculated for the two half-groups. The coefficient
of correlation thus obtained is .8221 +.0316. Prediction by the
Spearman-Brown prophecy formula 23 gives a probable correla-
tion of .9024+ .0181 between two groups, each of which is simi-
lar in size to the entire group of "poorest" readers.

*s The Spearman-Brown formula is stated;

(Holzinger, Statistical Methods for Students in Education^ p. 169.)

ELEMENTS INFLUENCING READING DIFFICULTY 127

Evidence has been frequently adduced which points to the
variability among reliability coefficients obtained by the split-
halves technique. In view of this evidence, the group was again
split into random halves and a second index of reliability com-
puted, as before. This time, a correlation of .9254^.0140 was
found between the half-groups, and a probable correlation of
.9613 .0074 by application of the Spearman-Brown formula.
Since the difference between the two coefficients, .0589, is sig-
nificant by the accepted test of significant differences, it seems
likely that neither can be taken as an absolute index of the re-
liability of the group sampled. Furthermore, there is no reason
to assume that the extremes of the possible range of coefficients
are represented by .9024 and .9613. The homogeneity of the
group so far as reading ability is concerned leads to the assump-
tion, however, that either coefficient may be taken as reasonably
reliable. The size of either coefficient seems large enough to jus-
tify accepting the average reading scores attained by 190 se-
lected readers as fairly reliable indexes of the difficulty which
would be experienced by a larger group of adults of similar
reading ability.

Table XXVII shows that the factor most closely correlated
with difficulty for "poorest" readers is the number of words in
a selection not known to 90 per cent of sixth-grade pupils. The
factor next most closely related to difficulty is the number of
different hard words in a selection. Further, the percentage of
easy words and the number of easy words in a selection are
about as closely correlated with difficulty,, although in an oppo-
site direction. Some of the factors show little relation to diffi-
culty of comprehension. If the selections studied are typical
of the general reading material of adults, these factors do not
generally influence its difficulty.

Table XXVII also shows that twenty-four expressional ele-
ments are significantly correlated with difficulty of general read-
ing materials of the sort used in the test items, that is, of gen-
eral magazines, newspapers, and books. Each of these elements
may, therefore, be taken as an indicator of the difficulty of

128

WHAT MAKES A BOOK READABLE

similar material, when difficulty is measured by the occurrence
of structural elements. Furthermore, a combination of all ele-
ments may be expected to give the best prediction of difficulty.
But, obviously, twenty-four are too many to take into account,

TABLE XXXI

CORRELATIONS OP SIGNIFICANT ELEMENTS OF DIFFICULTY WITH
AVERAGE READING SCORE AND WITH EACH OTHER

READ-
ING
SCORE

SIGNIFICANT ELEMENTS OF DIFFICULTY*

Read ing Score

I. OOO

- 550

--S43

.529

.518

.488

-.457

450

-449

-.441

-.428

.419

Elements*
i

.876

-.804
-.963

-.847
-.890
952

-.833
-.819
.836
.804

.660
-SS8
-.573

=:S?

- 713
- 539
612
593

593
- 659

-485

-:S

i:S
.910
-.640

424
.598
-.559
-.585
-.399
.171
-.190
.076

.801
-740
-743
.702
-.861
.687

349

-.463
.362
.421
439

-:g

-.434

|.. .......

18 ..

24.

Mean

0.87
0-57

9-83

6.21

26.12
9.28

74-06
8.07

83.81
10.31

70,92
6.99

27.31
11.85

9-54
5.96

19.24
7.09

64.73
5.06

10.67
6.26

6.50
2.6l

S.D

* Significant elements of difficulty:

1. Number of words not known to 90 per cent of sixth-grade pupils.

2. Number of different hard words.

3. Percentage of easy words.

4. Number of easy words.

5. Percentage of monosyllables.

6. Average sentence-length in syllables.

7. Number of firat-, second-, and third-person pronouns.

8. Average sentence-length in words.

9. Perceatage of different words.

10. Perceatage of polysyllables.

11. Number of explicit sentences.

especially since the most significant elements frequently meas-
ure approximately the same thing. For example, number of easy
words and percentage of easy words are merely two ways of
measuring the same element. Similarly, number of words not

ELEMENTS INFLUENCING READING DIFFICULTY 129

known to 90 per cent of sixth-grade pupils and number of dif-
ferent hard words both measure hardness of vocabulary. In such
cases it is apparent that the number may be reduced by using
only one of two closely related elements.

The usual method of discovering how much relationship exists
among a group of elements is that of intercorrelation. By this

TABLE XXXI Continued

SIGNIFICANT ELEMENTS OF DIFFICULTY

-.409

-.407

-.406

-.377

371

-.370

-.367

-307

-300

.298

-.285

-.279

-.271

877

.668
.719
-.708
-.685
555

" : f
III

--4S9

-:j

=:!

.810
-.632
.794

.2X8

577
-.703
.478

.414
.294
-.287
-.276
-.404
.809
-.523

-:JSS

.380

.413
.442
-.387

= I

-.304
.173
-054
.350

.112
.284

437
--3I5
375
.4"
411

.810
-.041
-.414
-977

.506
.516
-.509
-.467
-.499

-:%
:S8

.429
-.5"
.405

.404
.603
-.507
-.408
-37|
.188
-.143

7 i
.896

354
.015

.310

-.454
-.380
39S
431
.400
-.402
.556
-.390
-.327
-340
319
.340

-.179
-.275
.262
.241
.136
-.171
.087

-135
-.203
-.115
.000
.039

-.240
-.187
.223
.209
.247
-.692
.427
--775
-.043
-.309
.824
-.268

-094
.088
--OSI

-.073
.206

.167

.150

.188
.240
.156

.092
.075

.565
.612
-593
-.521

-.82

-.510
550
.225
SIS

"IS

-458
433
-.431
-.395
-.467
.440
-455
40S
035
.402
-.S3J
.406

.552

JflT

g .

.267

- 426

.003

.550

.899

,463

O48

666

?8T

.060

.334

.173

.721

276

.195

.227

"*"

- 088

443

.128

- 048

.335

.034

068

389

334

.501

.080

297

.026

777

-.086

-.468

530

.069

*99

.283

-.266

.325

.411

933

.341

.140

.023

.204

.299

.019

.176

.155

""243

342

.141

.059

.US

.060

.082

.119

3*7

.255

.043

.294

.450

33-56
17-47

49-00
19.23

12.04

8.12

0.81
1.32

6-75
3.09

11.38
3.58

73-79
S-7&

13.21
3.45

2.92
2.00

4.08
r.73

37-21
16.42

13.21
3-43

12.

13-
14.

JI:
II:

19-
20.

21.
22.
23-
24.

Percentage of different words not known to 90 per cent of sixth-grade pupils.

Maximum syllabic sentence-length.

Minimum syllabic sentence-length.

Number of figures of speech.

Number of sentences per paragraph.

Number of prepositional phrases..

Number of different words.

Number of third-person pronouns*

Number of infinitive phrases.

Number of simple sentences.

Number of * words

Range of syllabic sentence-4ength.
Number of infiniti'

.ve and prepositional phrases.

method it is possible to select for further use elements that cor-
relate as closely as possible with the criterion in this instance,
the average reading score of 190 readers and as little as possi-
ble with each other. Table XXXI shows the intercorrelation

130 WHAT MAKES A BOOK READABLE

of the twenty-four elements identified as significant indexes of
difficulty for "poorest" readers. 24

SELECTION OF MOST USEFUL ELEMENTS

Several criteria were set up in selecting useful elements for
prediction. It was believed that such elements should correlate
closely with the average reading score, correlate relatively little
among themselves, be readily recognizable, and together give an
adequate representation of known indexes of difficulty without
including a number too large for convenient use*

Eight elements seem to meet the requirements set up for se-
lection. They are: number of different hard words; number of
easy words; percentage of monosyllables; number of first-, sec-
ond-, and third-person pronouns; average sentence-length in
words; percentage of different words; number of prepositional
phrases ; and number of simple sentences. Examination of Table
XXXI justifies the selection of these elements from the point
of view of the first two criteria of usefulness.

With respect to the third criterion, one may ask whether dif-
ferent "hard" words, for example, are more readily recognized
than words "not known to 90 per cent of sixth-grade pupils/'
The answer is clearly, "yes." A "hard" word is any word that
is not "easy." And easy words, as they are defined in this study,
number only 756. They can be memorized, therefore, as they
are used. The result is that a reader is soon able to check hard
words with surprising rapidity.

Counting the number of words not known to 90 per cent of
sixth-grade pupils cannot be done so expeditiously. In the first
place, Dale's 8,ooo-Word List is distributed privately and hence
is less accessible than the Easy-Word List. In the second place,
no one can hope to remember whether a particular word is
known, let us say, by 42 per cent or by 92, per cent of sixth-
grade pupils. Every word must be checked against the list in
order that its familiarity may be determined. Furthermore, the

3 * In Appendix E, Table LXXXIV, 5s shown the Intel-correlation of the forty-four
elements described earlier in the Chapter.

ELEMENTS INFLUENCING READING DIFFICULTY 131

list obviously is not exhaustive, and a word may not appear
therein. It becomes necessary in such cases to resort to an esti-
mate of familiarity which may or may not be reliable. Number
of words not known to 90 per cent of sixth-grade pupils was
therefore considered less useful as a measure of hardness than
number of hard words.

Preference was given to word-counts as a measure of length
of sentence, for the reason that they can usually be made more
rapidly and accurately than counts by syllables. One element
of difficulty,, namely, figures of speech, warrants selection from a
statistical point of view, inasmuch as it correlates fairly closely
with the criterion of difficulty and relatively little with other
elements. However, it has not been selected among elements of
greatest usefulness. There seems reason to believe that unless a
person is sensitive to figurative style, he probably will err in
his count of figures of speech and hence obtain an unreliable
measure.

This sorting of elements of difficulty in no way precludes the
use of other significant elements, should one prefer to use them.
For greatest accuracy, however, elements which measure differ-
ent aspects of structural expression are most desirable. These
are the ones which we have just pointed out.

HOW SHALL THE MOST USEFUL ELEMENTS OF DIFFICULTY

BE USED IN ESTIMATING DIFFICULTY OF

READING MATERIALS?

In view of the fact that each of the eight elements selected in
the previous section is related to difficulty, each may be used as
an index of the difficulty of material in which it occurs. Further-
more, the occurrence of all or several elements of difficulty may
be considered a better index of relative difficulty than any single
element. For example, if in counting the number of different
hard words in equal sized samplings of two books, The Epic of
America and The Good Earth> let us say, we find, as is actually
the case, a8 per cent in the former and 15 per cent in the latter,
we have some grounds for thinking that The Epic of America

132 WHAT MAKES A BOOK READABLE

is the more difficult book. If we find also that 67 per cent of the
words in the first are different words as against 65 per cent in
the second, we have slightly more evidence. If we pursue the
analysis still farther and discover in The Epic of America an
average of 15 prepositional phrases in a sample similar in size to
one containing 8 prepositional phrases in The Good Earthy there
is added reason for ranking the first as more difficult. An aver-
age sentence-length of 44 words in the former also indicates
greater difficulty than an average length of 29 words in the lat-
ter. If we look again at elements which indicate ease and find
only 5 personal pronouns in the one and 13 in the other, we have
little reason to doubt the relative difficulty of the two books.
Although simple sentences, monosyllables, and number of easy
words might be included in the analysis, their use is made un-
necessary by the agreement among the elements already count-
ed. We already have sufficient evidence to rank The Epic of
America of greater difficulty than The Good Earth, since it con-
tains more different hard words, a larger percentage of different
words, fewer personal pronouns, more prepositional phrases,
and longer sentences.

The number of elements to be used in estimating difficulty
depends upon several factors: the degree of agreement reached
by the use of two* or three elements, the amount of time neces-
sary in making the estimate, the degree of precision desired, and
so on. Although elements do not always vary in the same direc-
tion with the consistency of the foregoing illustration, simplicity
or complexity of expression tends to be more or less general for
any particular book. All the eight elements which we have
designated as most useful seldom modify materially the esti-
mate of difficulty that can be made by four or five of their num-
ber. Proof of this statement will be shown later.

In estimating the relative difficulty of magazines, newspapers,
and books, as reported in the next chapter, four elements have
been used: number of monosyllables, percentage of different
words, average sentence-length in syllables, and number of
simple sentences. In some cases additional elements have been

ELEMENTS INFLUENCING READING DIFFICULTY 133

counted for the purpose of comparing the classification of ma-
terials by few or many elements of difficulty. These are: num-
ber of different hard words, percentage of easy words, number
of prepositional phrases, and number of personal pronouns.
Agreement among three analyses has been generally found to be
reliable.

One of the chief merits of the method just described probably
lies in its objectivity. Its use obviates the haphazard designa-
tion of a particular book, magazine, or newspaper as relatively
easy or difficult on the basis of personal opinion. A second merit
lies in its reliability, which was discovered by correlating the
absolute difficulty of a series of selections expressed in terms of
test scores with the relative difficulty of the same selection in-
dicated by rankings on the test. For fiction, a rank correlation
of .535 .103 was obtained; for non-fiction, ,663.o8i. Rank-
ing of materials for difficulty by this method is therefore fairly
reliable.

Two major limitations attend the ranking of materials ac-
cording to the number of elements of structural difficulty. To
return to our previous illustration. Let us suppose that after
ranking The Epic of America as more difficult than The Good
Earth we are asked to isolate one or the other and to define its
difficulty precisely. We discover, then, that although we know
how many different hard words the book contains, what per-
centage of the words are different, how many personal pronouns
it averages per hundred words, and the number or percentage
of several other elements, this is all we do know. Rather, this is
all, except that the book is more or less difficult than another
book which we have used as a basis of comparison. Some other
technique must be introduced in order to define the difficulty of
The Epic of America^ The Good Earth, or any book in terms
of an accepted criterion of difficulty. Such a technique is de-
scribed in the next section.

By the method just presented, we assume that all of the vari-
ous elements by which we estimate difficulty have equal
weights. For example, a high percentage of different hard words

134 WHAT MAKES A BOOK READABLE

and a long average sentence are considered equal contributors
to difficulty, In reality, such is not the case, since no group of
variables in a realm of concomitants acts in this manner. Hence,
the estimate is somewhat less exact than if each element were
assigned its proper weight, as is done by the method described
in the following paragraphs.

PREDICTING DIFFICULTY IN TERMS OF AN
AVERAGE READING SCORE

Since all elements of difficulty must serve in some way to de-
termine the difficulty of a selection in which they occur, it seems
probable that a combination of the most useful elements into a
single instrument of prediction will give the best possible esti-
mate of difficulty. That instrument is a statistical device known
as a "regression equation/' which expresses the relation between
a single dependent variable, X 19 and a number of independent
variables, X*, X 3 , X+ . . . . X n . In this case, X* is the criterion
of difficulty which we are trying to predict. That is, it is the
average reading score which a group of readers of limited ability
will probably make when tested on a selection. X 2 . . . . X n are
the elements of difficulty combined to predict JTi. The regres-
sion equation which may be used to predict X x is: 25

X*** - . 06566 JT a +. ooi 268J*VK 004064^4+
. 007 j45.Xi - . 02342^ - . 03371X7 -
. 01455^8-. oioi5-ST 9 +3. 408+ .2941 .

The meanings of the terms are as follows:

Jx=the average reading score which will probably be made by a group of
adults of limited reading ability on a given passage of general reading
material.

-XT~the number of different hard words found in a passage of the size used in
the reading tests., i.e., about 100 words in length.

-X" 3 =the number of easy words found in a passage of 100 words.

-3T 4 =the percentage of polysyllables found in the passage.

a s The general equation form for n variables is:

(Holzitiger, op. cit., Formula 139, p. 292.)

ELEMENTS INFLUENCING READING DIFFICULTY 135

X s = th& number of first-, second-, and third-person pronouns occurring in the

passage.

Jf6=the average sentence-length in words used in the passage.
X 7 =tht percentage of different words used in the passage.
Jf8=the number of prepositional phrases found in the passage.
Jf 9 =the number of simple sentences used in the passage.
.006566, +.0012685 etc., = regression coefficients which give the weight or

value to be attached to each independent variable.
3 .408 = a statistical constant.
. 2941 = the probable error of prediction.

The average reading score which will probably be made by
the group, when predicted by this equation, has a probable error
of .294. This means the chances are about even that the pre-
dicted score will not differ from the actual score which would be
obtained by testing more than ,294.

To illustrate the use of this equation in predicting the average
reading score of a selection, take, for example, the adapted ver-
sion of Robinson Crusoe. Samplings of this book contain on the
average 1 1 different hard words, 70 easy words, 72 per cent of
monosyllables, 16 personal pronouns, an average sentence-
length of 12.04 words, 36.5 per cent of different words, 7 prepo-
sitional phrases, and 4 simple sentences. The most probable
average score that readers of limited ability would make if they
were tested for comprehension of the book is obtained in the
following manner:

^ I =-(.oo6566Xn)+(.ooi268X7o)+(. 004064X72)
+(.oo75 4 5Xi3)-(. 02342X12. 04) -(.03371X36. 5)
~(. 01455 X7)~(- 01015X4)4-3- 408=2, i.

This average score of 2.1 indicates that readers of limited
ability will probably find the adapted Robinson Crusoe as easy
to read as any item used in the adult reading tests, since the
easiest item had an average score of 2.1 for "poorest" readers.
An examination of all individual scores indicates that 58.6 per
cent of the "poorest" readers showed a fair comprehension of
the selections by a score of 2 or better. A close relationship
exists between the percentage who can be expected to compre-
hend the selection and the average comprehension score which

136 WHAT MAKES A BOOK READABLE

the group can be expected to make when tested on the selection.
This has been shown by a correlation of .898. It seems probable,
therefore, that most adults of limited reading ability can under-
stand a selection with an index of difficulty of 2.1. What this
index means in terms of grade-levels will be shown in a later
chapter.

RELIABILITY OF PREDICTION

How well do the eight elements predict the difficulty of gen-
eral reading material? How much better than mere chance is
prediction by the technique described in the foregoing sections?
In other words, is prediction by the use of eight elements re-
liable? These questions may be answered in either of two ways.
According to the one, we may simply state that the eight ele-
ments predict difficulty fairly well and that their use gives an
estimate considerably better than mere chance. For the gen-
eral reader of this report, such an answer is probably satisfac-
tory. He can be content "to take our word" as truth and pass
over the remainder of this section. There we answer the ques-
tions in a less general way. Such an answer is intended for the
more analytical reader the student of research who wants
statistical proof that a proposed technique is reliable.

Precision of prediction depends partly on the size of the corre-
lation between the right and left sides of the equation, that is,
on the correlation between the actual and the predicted difficul-
ty of a selection. This relationship is expressed in terms of a
multiple-correlation coefficient, designated R. For the equation
given on page 134, R has a value of .64^^ It is evident that the
combination of eight elements yields a higher correlation with
difficulty than any of the elements taken alone. The correlation
of the single elements with difficulty has been shown in Table
XXVII.

From R, .645^ it is possible to measure the relationship be-

26 The formula for obtaining R is: Rifo n) *l : g ' ' ' '

(Holzinger, op. cit., Formula 155, p. 307.)

ELEMENTS INFLUENCING READING DIFFICULTY 137

tween prediction by the regression equation and a pure guess. 27
This relation was found to be .764, which means that the esti-
mate of difficulty based on the occurrence of eight elements will
be in error on the average by about .76 as much as if the errors
resulted from pure guesses. Or subtracting .76 from i.oo, the
errors will be .24 smaller than those in pure guesses.

Precision of prediction depends also on the size of the prob-
able error of estimate, commonly designated P-E.^.^,)- For
the equation developed in this section the value of the probable
error of estimate is .ap. 28 This means that the chances are about
even that the actual difficulty of a selection will not vary from
the predicted difficulty more than + .29. If the predicted diffi-
culty of a given book is represented by an average score of 2,1,
the chances are 50 in 100 that the actual difficulty will lie some-
where between 1.81 and 2,39. Although the relationship be-
tween prediction and actuality is considerably lower than an
approximation of certainty, it is higher than other investigators
have obtained in predicting difficulty of other types of material.
It is unquestionably high enough to be of practical value in
estimating difficulty of materials for adults of limited reading
ability.

PREDICTING DIFFICULTY BY A SMALLER
NUMBER OF ELEMENTS

In view of the amount of labor involved in counting the oc-
currence of eight elements in a selection, we made an effort to
discover whether combinations of fewer elements would give a
prediction approximately as good. One who is accustomed to
deal with statistical data will see at once that low weights are
attached to three of the variables in the equation number of
easy words, percentage of monosyllables, and number of simple

a ? The coefficient of alienation is calculated:

Truman L. Kelley, Statistical Method. (New York: Macmillan Co., 1923), p. 173.

28 The formula for the probable error of estimate is: P..( es t.) ==s .6745 <n .23 ..*
(Holzinger, op. cit. t Formula 84, p. an.)

138 WHAT MAKES A BOOK READABLE

sentences. Apparently these variables do not make for greater
predictive accuracy. When they are omitted from calculation
the new equation becomes:

jYjs .01029X2+ .009012X5 .02094X6 .03313X7
-.01485X8+3-774,

in which the subscripts of X designate the same elements as in
the longer equation. X 2 designates number of different hard
words; X s , number of first-, second-, and third-person pronouns;
Xe, average sentence-length in words; X 7 , percentage of differ-
ent words; and X 8) number of prepositional phrases. Quite as
reliable results may be obtained by combining these elements as
by combining with them the three omitted from the earlier
equation. The statistical proof for this statement lies in the fol-
lowing measures:

^==.6435+. 085
P.E. (est. X x )=.294

Multiple r, .6435, is almost identical with .6446 obtained for the
longer equation. The probable errors of estimate vary from
,2945 to .2941 for the two equations. This variation is clearly
insignificant.

By arranging the eight most useful elements into various
combinations of four, it has been found that nine different com-
binations will each give an estimate of difficulty about as good
as that obtained by the use of more elements. Equations for
these combinations are shown in Table XXXIL The symbols
have the same meaning as in the first equation on page 134.
Although a slightly better estimate of difficulty can be secured
from the use of one team of variables than another, there seems
to be no special advantage in giving preference to any particular
one, except on the ground of ease of counting. The formula used
in the chapters which follow is for variables 1.25678.

In order that estimates by the regression equation shall be
reliable, certain cautions need to be observed. Data to which
the regression equation is applied must be comparable with
those of the sample from which the equation was derived. In

ELEMENTS INFLUENCING READING DIFFICULTY 139

other words, they must be obtained from general reading ma-
terials of the sort used in the tests described in chapter iii.
Samples of the material should be approximately the same
length as the test items about 100 words. In predicting the
difficulty of long selections, several hundred-word passages may
be selected and examined for the presence of the elements of
prediction. An average of the occurrence of these elements in

TABLE XXXII
INSTRUMENTS OF PREDICTION, COMBINING FOUR ELEMENTS OF DIFFICULTY*

Vari-
ables

Formula

P.E.

(est.
-Xi)

1.2678..

X x = .01229X3 -.02448X5 .03245X7-. 01623X8+3. 952

.6402

.2956

.2567..

X z = .01333X2 +.01007X5 .02224X5-. 03054X7+3. 532

6393

.2958

.3678..

Xi= .009709X3 .02510X5 .03410X7 .01882X3+2.965

.6387

.2960

.4678..

Xi= .01279X4 .02405X5 .03866X7 .01962X8+3.156

-6385

.2961

.2467..

X x = .01156X3 +.006938X4- .02552X5- .03055X7+3 .153

.6368

.2967

.2679..

Xi= .01540X2 .02924X5 .02999X7 .01188X9+3.816

6355

.2970

.4567--

Xi= .01350X4 +.01057X5 .02247X5 .03758X7+2.682

-6347

3467--

Xx= .007945X3+ .009218X4 .02622X5 .03245X7+2 . 160

6345

.2974

35 6 7--

Xx= .01027X3 +.008694X3- .02437X5-. 03298X7+2. 535

.6340

-2975

* Xt** average reading score.

Jfa=number of different hard words in a passage of 100 words.
Jfj= number of easy words.
JS^ percentage of monosyllables.
Jfssanuraber of personal pronouns.
X & = average sentence-length in words.
JT7=percentage of different words.
Jfjas number of prepositional phrases.
J$r 9 =number of simple sentences.

the selected passages should then be substituted in the regres-
sion equation. Finally, the estimated score must be interpreted
as the difficulty which adults of limited reading ability probably
will experience in reading.

VERIFYING THE PREDICTED SCORES

Is a formula derived from the scores of one population a valid
means of predicting scores for a second population ? The pro-
cedure followed in answering this question was, first, to obtain
measures of reading ability of a second population by use of the
Adult Reading Test, and then to correlate the scores with the

140 WHAT MAKES A BOOK READABLE

predicted scores based on the performance of the first popula-
tion.

One hundred and thirty-six adults in evening classes and in an
industrial reformatory were tested. Of this number, 102 had
scores that fell within the limits which identified "poorest" read-
ers, that is, between 68.63 and 15.05. The scores obtained by
these readers were then correlated with those predicted by use
of the regression equation based on the scores of 190 "poorest"
readers. A coefficient of .720 .07 5 indicates that the regres-
sion equation is valid for use in a second population of which the
experimental population may be taken as an adequate sample.

SAMPLING OF BOOKS

The statement has already been made that the elements used
in the regression equation should be counted in passages ap-
proximately one hundred words in length; also that the counts
obtained from several such passages in a long selection should
be averaged to obtain the typical occurrence of particular ele-
ments in the entire selection. In interpreting the average of sev-
eral passages as characteristic of the larger selection, we are
faced with questions concerning the adequacy and the repre-
sentativeness of the sampling.

The question of adequacy of the sampling relates to the num-
ber of hundred-word passages to analyze in a given selection.
The question of representativeness pertains to the distribution
of the samplings, so that they will represent fairly well all por-
tions of the selection. The procedure commonly recommended
to research workers is to exercise best judgment and expert
knowledge of the situation in choosing what shall constitute a
reliable sample and to select the samples in a way that appears
to be fairly representative. For all practical purposes, a sample
paragraph of approximately one hundred words from each chap-
ter seems adequate. If examination of a book indicates that
chapter divisions are more a convenience of format than a logi-
cal division of thought, then one passage may be taken as rep-
resentative of several chapters.

ELEMENTS INFLUENCING READING DIFFICULTY 141

Some evidence of the representativeness of the hundred-word
sampling of each chapter is shown in Table XXXIIL The data
present the average occurrence of the five elements of difficulty
in two hundred-word passages from each of twenty chapters in
Everyday Problems of the Everyday Child. Differences in the
mean occurrence of each element in the two passages are not
statistically significant. For this book, then, one sampling is as

TABLE XXXIII

AVERAGE OCCURRENCE OF FIVE ELEMENTS IN Two SAMPLINGS OF TWENTY
PASSAGES EACH FROM ONE BOOK

ELEMENT*

FIRST SAMPLING OF
TWENTY PASSAGES

SECOND SAMPLING OF
TWENTY PASSAGES

DIFFERENCE

BETWEEN

MEANS

Average

S.D.

Average

S.D.

29-6o .6819
4-65+ .3324

30-7211-590
yo.32 .6771

11.65+ -4 110

4.521
2.204
10.54
4.489
2.725

29.00+ .7401

5-05 -5342
a8.53i. 49 o
68.34 .6888
i2,45 -559 1

4.907

3-542
9.868

4-5*7
3.707

. 60 + 1 . 01 1
.40 .6291
i.i92.i79
i-98 .9648
.8o .6939

6 ...

* Elements:
a. Number of different hard words.

5. Number of first-, second-^ and third-person pronouns.

6. Average sentence-length m words.

7. Percentage of different words.

8. Number of prepositional phrases.

representative of structural elements in a chapter as two sam-
plings. While a larger number of passages would probably give
a more precise sampling of the entire content, it is scarcely prac-
tical to use them in view of the time required to analyze each.

SUMMARY

The chapter has presented partial answers to two questions.
The first is, What elements of difficulty are inherent in adult
reading materials of a general nature? The second is> How can
these elements be used in a more accurate estimate of the diffi-
culty of general materials for adults of limited reading ability?

In answering the first question, elements of difficulty have

14* WHAT MAKES A BOOK READABLE

been defined for adult readers in general, for "best" readers,
and for "poorest" readers. The facts show that while an ele-
ment of difficulty tends to be general, the extent of its influence
tends to vary with specific classes of readers. The size of the
coefficients and direction of their deviation indicates that as
reading ability reaches a higher level of efficiency, difficulty of
comprehension, due to variations in the author's expression,
tends to grow less. Although the present chapter has gone no
farther than to define good and poor readers in terms of their
average performance on the Adult Reading Test, it seems rea-
sonable to assume that directors and advisers of adult groups
can estimate with considerable accuracy the class in which par-
ticular individuals belong.

In answer to the second question, two methods of estimating
difficulty have been presented. The first involves a compara-
tive count of significant elements of difficulty occurring in dif-
ferent reading materials. It is a simple and fairly reliable means
of judging the relative difficulty of books, magazines, and news-
papers. The second supplies a more specific and somewhat more
reliable method of predicting difficulty in terms of the average
reading score. The data presented show that a regression equa-
tion involving eight elements makes for accuracy of prediction
little better than that obtained by combinations of fewer ele-
ments. Although techniques developed in this study do not
yield highly accurate estimates of difficulty, predictions based
upon a reliability of .64 will be fairly satisfactory for general
prediction and unquestionably more effective than sheer chance.

Succeeding chapters illustrate how the findings presented
here may be put to practical use by librarians and others. Some-
times they are used to estimate in an objective and reasonably
reliable manner the relative difficulty of general magazines and
newspapers. Again, they serve as a means of predicting absolute
difficulty in terms of an average reading score.

CHAPTER V

HOW DO ADULT READING MATERIALS DIFFER

WITH RESPECT TO ELEMENTS

OF DIFFICULTY?

IT IS not within the scope of this report to analyze any large
amount of adult reading material. It seems essential, how-
ever, to examine a considerable sampling of several kinds
for the purpose of answering such questions as: How do ma-
terials differ with respect to significant elements of difficulty?
What materials can be designated relatively easy or difficult on
the basis of one or more elements of difficulty inherent in them ?
In answering these questions, we have analyzed representa-
tive magazines, books, and newspapers to discover the occur-
rence of elements that characterize them as relatively easy or
difficult. Four elements which correlate significantly with diffi-
culty have been used: percentage of monosyllables; percentage
of different words, or vocabulary diversity; length of sentence
in syllables; and number of simple sentences. Any conclusions
concerning relative difficulty are limited to these elements.

WHAT MATERIALS TO STUDY

One of the most perplexing problems connected with a study
of reading matter is the selection of materials that may claim
to be representative of what is written for adults and presum-
ably read by them. If at least half the adult public reads maga-
zines, newspapers, and books, then these three sources of ma-
terial may be thought of as generally representative of what
adults read. But they suggest only a broad classification. They
lack particularity. Obviously, a book is not just a book. Nor is
a magazine just a magazine. Each has some mark of individu-
ality, despite the notable standardization of content, style, and
format that prevails among certain classes of newspapers, popu-

143

144 WHAT MAKES A BOOK READABLE

lar magazines, and best-sellers. Selecting materials from the
three sources, then, involves, first, defining arbitrary criteria of
selection, then examining and sorting materials in the light of
these criteria, and, finally, assembling those which seem to meet
the criteria most satisfactorily.

THE MAGAZINES STUDIED

In order to keep the list of magazines within manageable pro-
portions, we have limited the study to an analysis of those which
are classified as "General" in the Directory of Newspapers and
Periodicals? They contain a wide variety of articles intended to
meet the needs of a heterogeneous public. Classifications ad-
dressed to specialized groups of readers are therefore omitted.
General magazines are further restricted to those whose circu-
lation figures equal or exceed 125,000, as sworn to by the Audit
Bureau of Circulation. If the circulation figures are a publisher's
claim, unsupported by affidavit or detailed statement, or are
merely estimates of another sort, the magazine is excluded.

According to Ayer, circulation means "the average number
of complete copies of all regular issues for a given period, ex-
clusive of left-over, returned, file, sample, exchange, or ad-
vertisers' copies, and special editions." 2 A list of magazines
based on circulation figures, therefore, indicates the best-selling
magazines and may be accepted as indicative of the preferences
of magazine purchasers. But since circulation is conditioned by
supply, which in any given community may be affected by the
adequacy of facilities characterizing its newsstands and by other
external influences, preference for magazines, as determined by
the number purchased, obviously may fall short of representing
the true number actually read.

The sixty-eight general magazines which comprise the list
cannot be said to be representative of the general reading of
adults except in so far as they presumably represent in every

1 N. W. Ayer & Sons, Directory of Newspapers and Periodicals for 1931 (Philadelphia:
N. W. Ayer & Sons, 1931), pp. 1250-55, 1256-59.

. 8.

HOW ADULT READING MATERIALS DIFFER. 145

TABLE XXXIV

NAMES AND DATES OF ISSUE OF MAGAZINES USED IN THE ANALYSIS
AND THE ANNUAL CIRCULATION OF EACH

Name of Magazine

Issue

Circulation

Ace-High Magazine

March, 1931

I CQ. 820

Aces (Fiction House Group)

March, 1031

Sl^.Al?

All-Star Detective Stories

March, 1931

706 . ci i

All-Story

March 21, 1931

V7Q . {-76

American Home

February IQTI

2 CO . 1 7 <

American Magazine, The

February. IQ^I

o 070.108

Atlantic Monthly, The

February. IQTI

I2q.7o8

Better Homes and Gardens

February, IQII

* 7 / 7 U

I , 7QO , 660

Blue Book Magazine, The

March, 1931

I7C,,7OI

Capper's Farmer

January, 1071

0^7 . A-A.A.

Clues

April. IQ*?I

211 .A^-d

College Humor . ....

February. IQTI

24.O , Q7 C

Collier's Magazine

February 1 IQTI

2 2C7 2QO

Country Gentleman

February, IOTI

I ,7OI ^QQ

Delineator . .

February icni

2 OO2.672

Detective Story Magazine

March 28, 1931

I,i47,c8o

Dream World

April. 1 07 1

210,74.0

Farmer's Wife *

February. lo^i

n^Q.OCC

Field and Stream

ApriL IQ^I

I74.,OQ2

Gentlewoman

December, 1930

1,158.204

Golden Book

January, 10^1

136,584

Good Housekeeping

February, 10.11

1,767.380

Hearst's International Cosmopolitan
Magazine

February, 1031

I 9 590 , 840

Holiday

March, 1931

153*3^6

Holland's Magazine ....

April, IQTI

414,111

House and Garden

March, 1931

I25,8l8

Household Magazine

February, IQ^I

1,777,088

Tudee

February 21, 1931

171,898

Ladies* Horne Journal, THe - - -

February. lo^i

2 1 <8i,g4a

Liberty

February 7, 1931

2,415,942

Literary Digest

February 14, 1931

1,602,397

McCall's Magazine

January, IQII

2,505,088

Mother's Home Life

February, 1931

723,669

Motion Picture Classic,

March, 1931

159,600

IVtotion Picture Magazine. . ,,..,..

April, iq^i

387,396

Nation's Business

March, 1931

312,076

National Farm Journal

December, 1930

1,517,446

146

WHAT MAKES A BOOK READABLE

TABLE XXXWConiinuec?

Name of Magazine

Issue

Circulation

National Geographic Magazine

Fe bruary 1 93 1

i ,2,01 ,082

Needlecraft

February, icni

I ,O27, ICQ

New Movie Magazine

February, IQTI

I . 2Q7 , 048

Outdoor Life

April i en i

I1C,4.W

Pathfinder

February 14., IQ'U

qq8 , Q4.8

People's Popular Monthly

February, 1931

I j 117 2 ?8

Photoplay Magazine

March, 1931

620,331

Pictorial Review . . ...

March, 1931

2,502,214

Picture Play

April, 1 93 1

2OC,OCQ

Popular Mechanics

March, 1931

514,810

Popular Science Monthly

April, 1 07 1

?6l ,OCQ

Radio News

March. 10*31

151 ,421

Ranch Romances

April, i on

211,434

Red Book Magazine

February. io?i

6^8,282

Review of Reviews . .

February. 10^1

i6o,ooc

Rotarian

March, 1931

1^8, 0^6

Saturday Evening Post

February 21, 1931

2,924,363

Science and Invention. .

April. IQ7I

14.8,4.^6

Screen Book Magazine . .

April, i o^i

260 . 20^

Screenland .... . ....

April. io? i

iSc.-ua

Short Stories

April, 1 07 1

I 'to, 009

Sky Riders

April, 1931

176,180

Successful Farming

December, 1930

I,IC7,8lI

Xime

February 16, 1931

<OO,I72

True Detective Mysteries

February. IQII

A86. 1 T7

True Romances

March, 1931

601 ,QCI

True Story Magazine

February* IQTI

2,110,^87

Vogue

March, 1931

I'JQ.QQI

Woman's Home Companion

January* icm

2,606.12^

Woman's World

March, icm

I.22C.7'?4

World's Work

March. IQ^I

J 27, 77 A

case the magazine reading of 125,000 readers or more. A list of
these magazines, with the dates of publication used, together
with circulation figures for 1931, appears in Table XXXIV. 3

THE NEWSPAPERS STUDIED

The Directory of Newspapers and Periodicals for 1933 reports
that 2,053 daily newspapers were published in the United States

*Ibtd., pp. 1203-11, 1212-18, 1250-55, 1256-59.

HOW ADULT READING MATERIALS DIFFER 147

and outlying territories in 1932. In the same year there were
10,760 weekly publications, of which some 9,000 were largely
newspapers of village and rural communities. The remainder
were trade publications of various kinds. Similar facts have
been compiled for the past several years, making it apparent
that if we are to study newspapers which represent what 95 per
cent of our population are reading, we should include daily and
weekly newspapers of city, village, and rural circulation. Fur-
thermore, the introduction of the graphic element into news-
papers, for the purpose of featuring news which even the illiter-
ate can understand, in a measure suggests the need of including
graphics or tabloids in the study.

In general, the ten daily papers studied are those having the
widest circulation among either morning or evening editions.
Three of these are graphics. Village newspapers represent week-
ly publications of towns in the Middle West having a popula-
tion ranging between 400 and 6,000. Rural publications are rep-
resented by Capper's Weekly. A total of fifteen newspapers com-
prise the list shown in Table XXXV.

BOOKS ANALYZED FOR RELATIVE DIFFICULTY

Inasmuch as the study has been restricted to an investigation
of elements of difficulty in adult reading material of a general
nature, only books of a general character are included. Books
on technical and vocational subjects, books that are encycloped-
ic or analytical in content, and books adapted to particularly
uncatholic tastes have not been included in the list. Those
which seem appropriately designated "general" are popular
novels, popular books of non-fiction, standard fiction, the Bible,
and certain miscellaneous types to be described later. Table
XXXVI lists the twenty-nine books which are ranked for rela-
tive difficulty in this chapter. Restrictions instituted in the
preparation of this list are admittedly arbitrary.

By popular novels and popular general books we mean those
that are currently popular, that is, books which have dominated
public interest since January, 1929. Authority for popularity

148

WHAT MAKES A BOOK READABLE

was found in demands at libraries and bookshops, as they were
reported in monthly book-lists in the Bookman and booksellers'
reports in Publishers' Weekly during the period from January,
1929, until June, 1931.

Inasmuch as standard books differ from popular books in
style and theme, it may be assumed that they differ also in the
occurrence of elements which mark a book easy or hard. Some

TABLE XXXV

NAMES or FIFTEEN NEWSPAPERS SELECTED FOR ANALYSIS, WITH CLASSIFI-
CATION AND DATE OF ISSUE OF EACH

Classification

Newspaper

Issue

Metropolitan

Boston Post
Chicago American
Chicago Daily Times (picture newspaper)
Chicago Daily Tribune
Christian Science Monitor

August 8, 1931
August 10, 1931
August 10, 1931
August n, 1931
August 8, 1931

Village

New York Evening Graphic
New York Journal
New York News (picture newspaper)
Los Angeles Times
Philadelphia Evening Bulletin

'Edmore Herald News (North Dakota)
Jefferson Banner (Wisconsin)

August 7, 1931
August 8, 1931
August 9, 1931
August 6, 1931
August 8, 1931

August 13, 1931
August 13, 1931

Rural Digest .

Olney Daily Mail (Illinois)
JRiceville Recorder (Iowa)

Capper's Weekly (Kansas)

June 27, 1931
August 5, 1931

February 7. IQ^I

support for this assumption is found in Scudder's comparative
study of length of sentences used by five nineteenth-century
and five contemporary writers. 4 His findings show that, with
the exception of Edith Wharton, all modern writers use shorter
sentences than did novelists of earlier periods. Hawthorne's sen-
tences, for example, average 36.42 words, as against sentences
with an average of 18.75 wor ds used by Willa Gather.
That the so-called "classics" seem to present no greater diffi-

- Harold H. Scudder, "Sentence Length," English Journal, XII (November, 1923),
617-20.

HOW ADULT READING MATERIALS DIFFER

TABLE XXXVI

NAMES OF TWENTY-NINE BOOKS SELECTED FOR ANALYSIS,
WITH CLASSIFICATION AND AUTHOR OF EACH

149

Classification

Book

Author

("All Quiet on the Western Front

Erich Maria Remarque

Angel Pavement

J. B. Priestley

Cimarron

Edna Ferber

Popular novels .

< Peder Victorious

O. E. Rolvaag

Roper's Row

Warwick Deeping

Scarlet Sister Mary

Julia Peterkin

[White Oaks of Jama

Mazo de la Roche

[Art of Thinking, The

Ernest Dimnet

Byron

Andre" Maurois

Elizabeth and Essex

Lytton Strachey

Popular general books. .

< Henry the Eighth

Francis Hackett

Preface to Morals, A
Story of San Michele, The

Walter Lippmann
Axel Munthe

Tragic Era, The

Claude G. Bowers

Adventures of Huckleberry

Finn, The

Mark Twain

"Best'* standard books .

Last of the Mohicans, The

James Fenimore Cooper

Scarlet Letter, The

Nathaniel Hawthorne

Tales

Edgar Allan Poe

[Moby Dick

Herbert Melville

Classics,, original form. .

< Robinson Crusoe

Daniel Defoe

[Silas Marner

George Eliot

[Moby Dick

Sylvia Chatfield Bates

Classics, adapted form. .

< Robinson Crusoe

Michael West

[Silas Marner

Ettie Lee

Adaptation in basic Eng-

lish

Carl and Anna

Frank Leonhard

School readers for mid-
dle grades

[Reading and Living, Book I
\ Reading and Living, Book II

Hill, Lyman, and Moore
Hill, Lyman, and Moore

(Reading and Living, Book III

Hill, Lyman, and Moore

Bible

King James Version

150 WHAT MAKES A BOOK READABLE

culty from the standpoint of vocabulary than popular modern
novels is shown by Witty and LaBrant. 5 They conclude from
their analysis that other elements than vocabulary must deter-
mine differences in difficulty, if such differences actually exist.
The standard books listed in Table XXXVI represent practical
unanimity of judgment among 400 professors of literature con-
cerning "the ten works by American writers that best represent
our bid for a permanent place among the masterpieces of the
world's literature." 6

Two versions of other standard books or classics are listed in
Table XXXVI . These are included for the purpose of ascertain-
ing how much simplification has been accomplished by adapting
the original books to low-reading levels and, if possible, what
method of simplification has been used.

The inclusion of a series of school readers for the middle
grades is based on the assumption that their content has char-
acteristics similar to adult books, since reading instruction in
the middle grades continues "until habits of rapid, silent read-
ing approach maturity/' 7 The series studied was selected at
random from texts published since 1929.

THE PROCEDURE FOLLOWED IN ESTIMATING
RELATIVE DIFFICULTY

The method used in estimating the relative difficulty of gen-
eral books, magazines, and newspapers has been described in
chapter iv. It involves the following: (i) analyzing the material
for the occurrence of four significant elements of difficulty,
namely, percentage of monosyllables, percentage of different
words, average length of sentences in syllables, and percentage
of simple sentences; (2) comparing the occurrence of these ele-
ments in the three classes of material taken as a whole; (3) com-

s Paul A. Witty and Lou L. LaBrant, "Vocabulary and Reading," School and Society,
XXXI (February 22, 1930), 268-72.

*H. W. L., "Million Books and Best Books," Golden Book, VIII (August, 1926),
382-83.

7 Twenty-fourth Year Book of the National Society for the Study of Education (Bloom-
ington, Illinois: Public School Publishing Co., 1925), Part I, pp. 55-56.

HOW ADULT READING MATERIALS DIFFER 151

paring the occurrence of the elements in materials within each
class; and (4) ranking materials within each class for relative
difficulty as defined by each element and by all elements.

TREATMENT OF FINDINGS

Striking similarities and differences were found in the analysis
of magazines, newspapers, and books. We have attempted to
show these findings in three ways. The first involves statistical
summaries of the number and proportionate occurrence of the
various elements in the different classes of material. The second
utilizes a combined statistical and graphical presentation of the
mean occurrence of each element in particular magazines, news-
papers, and books. And the third gives a classification of all
materials in each class according to their relative difficulty as
indicated by the percentage of monosyllables, percentage of
different words, length of sentence, and number of simple sen-
tences.

The first of these methods of treatment summarizes the find-
ings in a manner that shows dominant tendencies among differ-
ent types of material and makes possible certain generaliza-
tions concerning each. The second method is more detailed. It
shows similarities and differences among individual magazines,
newspapers, and books with respect to particular elements
studied. By isolating specific facts, it provides a means by
which materials can be ranked for difficulty. The third method
of treatment presents a classification of reading materials from
the easiest to the most difficult, thereby defining the area of
difficulty represented by each.

MONOSYLLABLES IN ADULT READING MATERIALS

Table XXXVII summarizes the data obtained from an analy-
sis of monosyllables in general magazines, newspapers, and
books. In all materials there is a predominance of one-syllable
words. The highest percentages are, respectively, 75.8 in maga-
zines, 70.6 in newspapers, and 83.3 in books; whereas the lowest
percentages are, respectively, 61.5, 61.3, and 61.9.

152

WHAT MAKES A BOOK READABLE

Some qualification should be made in interpreting the data
pertaining to books, for the reason that simplified books are
included in this class and therefore contribute to the statistical
summaries presented in the table. When simplified books are
excluded, the percentage of monosyllables is reduced from a
range between 83.3 and 61.9 per cent to a range between 77.6
and 61.9 per cent. The latter is not distinctly different from the
range of monosyllables found in general magazines.

TABLE XXXVII

SUMMARY OF DATA OBTAINED FROM AN ANALYSIS OF MONOSYLLABLES

IN THE READING MATERIAL OF MAGAZINES,

NEWSPAPERS, AND BOOKS

General
Magazines

Newspapers

Books

Range

75.8 -61.5
69.2 .28

3-37 - 20
69.2

7*'4
67-5
3-9
i-9

70.6 -61.3

64-3 -47
2.68 .33

64-3
65.4
62.0
3-4
i-7

83.3 -61.9
73-4 -55
4-4^ .39
73 - 2
75-9
69.5
6-4
3-a

Mean

Standard deviation

Median

Third quartile

First quartile

Quartile range

Quartile deviation

The closest agreement in percentage of monosyllables occurs
among newspapers. A difference of 9.3 per cent has been found
between the newspaper with highest percentage of monosyl-
lables, 70.6, and the one with the lowest percentage, 61*3. Such
consistency can probably be explained by the fact that several
newspapers of approximately the same date of issue tend to
present much news that is identical Further evidence of this
agreement is shown by the relatively low quartile deviation and
standard deviation,

A simple way to calculate the extent of variation among read-
ing materials with respect to the occurrence of monosyllables is
to apply the standard deviation to the mean. It is then evident
that approximately the middle two-thirds of all the material

HOW ADULT READING MATERIALS DIFFER 153

studied in general magazines contains between 72.6 per cent
(69.2+3.37) and 65,8 per cent (69.23.37) of monosyllables;
in newspapers, between 67.0 and 61.6 per cent; and in books, be-
tween 77.8 and 69.0 per cent.

Other facts regarding word-length may be obtained from the
medians and quartiles shown in Table XXXVII. These meas-
ures lead to the observation that in the middle half of all maga-
zines studied, the percentage of monosyllables varies from 71.1
(69.2+1.9) to 67.3 (69.2-1.9). In three-fourths of them, the
percentage of monosyllables exceeds 67.5 (j x ).

Conclusions of a similar sort may be drawn with respect to
other types of material. It is clear that all classes of material
when taken as a whole contain a high percentage of monosyl-
labic words. Newspapers, however, consistently contain longer
words than general magazines and books.

DIVERSITY OF VOCABULARY IN ADULT
READING MATERIALS

Table XXXVIII presents a summary of data obtained from
an analysis of materials for percentage of different words. A
comparison of statistical measures for the three types of ma-
terial reveals the following facts: The highest percentage of dif-
ferent words in a thousand, 55.9, is found in general magazines;
the lowest percentage, 25.5, in books. The exclusion of adapted
classics would still give books the lowest percentage. When the
three classes of material are compared in terms of the mean per-
centage of different words occurring in all material of a class, it
may be noted that books contain the lowest and newspapers the
highest percentage. These findings are to be expected. For al-
though both classes of material aim to make a universal appeal,
the former does so by presenting a single theme of interest,
whereas the latter utilizes a range of themes theoretically as
wide as the whole scale of human interest.

The range of different words is closest among newspapers. A
difference of only 7 per cent may be noted between the news-
paper containing the largest number of different words and the

WHAT MAKES A BOOK READABLE

newspaper containing the smallest. The general consistency in
percentage of different words for all classes of material agrees
with that found in percentage of monosyllables. As in the previ-
ous instance, the degree of spread is shown by quartile and
standard deviations, which are smallest for newspapers and
largest for books.

In material of books, the range extends from 50.6 per cent of
different words to 25.5 per cent when simplified books are in-
eluded, or to 36.1 per cent when they are excluded. In either

TABLE XXXVIII

SUMMARY OF DATA OBTAINED FROM AN ANALYSIS OF PERCENTAGE OF
DIFFERENT WORDS IN THE READING MATERIAL OF
MAGAZINES, NEWSPAPERS, AND BOOKS

General
Magazines

Newspapers

Books

Range

CC Q 44, O

<*5.2 4.6.2

CO. 6 2C.C

Mean. . . .

ro c + IQ

ro . o 4- .11

4.2 . C 4- .6q

Standard deviation

2.41+ .13

1. 77 4- .22

c.c8+ .40

Median

<O. C

ci .3

44.4

Third quartile

<2 I

C2.2

4.6. C

First Quartile

4.Q. 2

CO. 3

^Q.2

Ouartile ranffe ...

2 o

1 .0

7. 3

Quartile deviation

1.4.

1 .0

3-6

case, the range is several per cent wider than for other types of
material. The highest percentage of different words found in
any book is lower than in newspapers and general magazines.
We are led to conclude, therefore, that the books studied not
only contain a smaller percentage of different words than other
classes of material but they vary more widely in extent of vo-
cabulary. This conclusion is supported by other facts in the
table. For example, when we apply each standard deviation to
its respective mean, we find that in the middle two-thirds of the
general magazines studied, different words per thousand range
from 52.91 to 48.09 per cent; in newspapers, from 52.67 to
49.13 per cent; while in books the percentage of range for a
proportionate number of materials is between 48.08 and 36.92
per cent.

HOW ADULT READING MATERIALS DIFFER 155

LENGTH OF SENTENCES IN ADULT
READING MATERIALS

In analyzing the length of sentences sampled from maga-
zines, newspapers, and books, we have expressed length in
syllables rather than in words, for the measure had been used
previously in analyzing a part of the materials in an earlier
study and some data were already available. Since both meas-
ures have been found to correlate closely with difficulty, either
one or the other might have been used.

The general trends in sentence-length shown in Table
XXXIX are based on an analysis of 200 sentences in each of the

TABLE XXXIX

SUMMARY OF DATA OBTAINED FROM ANT ANALYSIS OF SYLLABIC SENTENCE-
LENGTH IN THE READING MATERIAL OF MAGAZINES,
NEWSPAPERS, AND BOOKS

General
Magazines

Newspapers

Books

Ranee . ....

202 i

IAQ i

2Q7 I

JVj?

.Median

21 . C

??.o

24.0

Third quartile

28.O

"?6.

'U.g

First quartile

IB 2

27 O

i6.c

Quartile range

Q.8

Q.O

1C. Q

Quartile deviation

4..Q

4. <

7.7

materials sampled. From this table several differences are out-
standing. Although all types of material contain sentences of
one word, no magazine or newspaper contains a sentence as long
as 297 syllables, found in one book, the original Robinson Crusoe.
Were this book excluded, the longest sentence, 186 syllables,
would then be comparable with the longest sentence in other
materials.

A comparison of medians shows that half of the sentences in
newspapers are at least 33 syllables in length; half of the sen-
tences in books are 24 syllables or more in length; and in gen-
eral magazines, half of the sentences equal or exceed 2,1.5 syl-
lables. From these facts and from the size of the third quartile,

156 WHAT MAKES A BOOK READABLE

it is evident that the newspapers and books studied contain a
larger proportion of long sentences than general magazines.

The agreement among materials of a class with respect to
sentence-length is closest among newspapers and least in books,
as shown by the quartile deviations. Again, the disparity among
books seems to reflect the influence of adapted classics whose
sentences have been intentionally shortened for the sake of
simplicity.

Since this table presents summarized data, it can indicate
only general trends of sentence-length and not the significant
variations that actually appear among sentences in individual
magazines, newspapers, and books. Some of these will be pre-
sented later.

SIMPLICITY OF SENTENCE STRUCTURE IN
ADULT READING MATERIALS

The facts to be discussed in this section are derived from data
pertaining to simple sentences which are summarized in Table
XL. If one recalls that there are five kinds of sentences com-
monly recognized in a classification of sentences according to
form simple, compound, complex, compound-complex, and
fragmentary then the mean percentage of simple sentences
suggests that this kind of sentence probably predominates in all
types of material studied. From an actual count of other kinds
of sentences, the results of which are not shown in this report,
we have evidence to the effect that this hypothesis is true only
for general magazines. In newspapers, complex sentences ex-
ceed simple sentences by 3 per cent. In books, they exceed sim-
ple sentences by about I per cent. Other kinds of sentences are
relatively infrequent in all classes of materials.

The main facts to be derived from Table XL include the fol-
lowing: The range shows that no newspaper contains more than
50 per cent of simple sentences, whereas magazines contain as
many as 62 per cent and books, 55.5 per cent. The extreme pre-
dominance of simple sentences, 62 per cent, occurs in a wood-
pulp magazine of cheap stories of action, adventure, and in-

HOW ADULT READING MATERIALS DIFFER 157

trigue. Writers of this sort of fiction uniformly adopt crisp, sim-
ple sentences or fragments of sentences in order to produce
startling effects.

The size of the quartile deviations shows that general maga-
zines follow the most consistent practice in use of simple sen-
tences; and books, the least. The latter circumstance proba-
bly is explained by the inclusion of adapted texts with simplified
sentence-structure, as well as the older classics in which the

TABLE XL

SUMMARY OF DATA OBTAINED FROM AN ANALYSIS OF SIMPLE SENTENCES

IN THE READING MATERIAL OF MAGAZINES,

NEWSPAPERS, AND BOOKS

General
Magazines

Newspapers

Books

Ranee

62.0 27.0
45- 6 -55

6.J2 .78

45-5
50.0
4 i.o
9.0
4-5

50.0 28.0
41.5 i. 12
6.42+ .79
42.0
47.0
37-

10.

55-5-4-0
35. 4 i. 4 6
n.7i.o3
37-
43-5
28.8

J 4-7
7-4

Mean

Standard, deviation

Median

Third quartile

First quartile

Quartile range . ...

Quartile deviation

sentences are notably involved. Greater uniformity might have
resulted from confining books to current publications, as was
done in the case of magazines and newspapers. Such a proce-
dure seemed impossible, however, owing to the very nature of
book-reading habits, which at their best are not restricted to
current material

DEFINING RELATIVE DIFFICULTY WITH RESPECT TO
SPECIFIC ELEMENTS OF DIFFICULTY

From the data presented in the previous section we have been
able to show what in general are the dominant characteristics
of magazines, newspapers, and books with respect to four ele-
ments of difficulty, and to compare in a broad way the different
classes of.reading material. But that is all. And it is not enough.

158 WHAT MAKES A BOOK READABLE

For we are interested not only in similarities and differences
among classes of reading materials, but in the variations within
each class. Is the proportion of monosyllables the same in At-
lantic Monthly as in Liberty** Does the reader of Capper 3 Week-
ly meet the same average sentence-lengths as the reader of
Christian Science Monitorl How much has the percentage of
simple sentences been increased in adapting Silas Marner for
low levels of ability? Such questions as these must be answered
in order to secure evidence by which a particular magazine,
newspaper, or book is designated more or less difficult than
another of its class.

The procedure followed in defining relative difficulty involves
two steps. The first is to discover variations in the occurrence
of the several elements of difficulty in individual materials with
respect to the mean or median occurrence in all materials of a
particular class. The second is, then, to classify materials into
three groups designated "easy," "average," or "difficult." Here,
"easy" is defined as an extreme deviation from the central tend-
ency for a particular element in the direction of simplicity.
"Difficult" is defined as extreme deviation in the opposite di-
rection, that is, toward complexity. And "average" is defined
as coincidence or approximate coincidence with the central
tendency.

GENERAL MAGAZINES CLASSIFIED ACCORDING TO
PERCENTAGE OF MONOSYLLABLES

Table XLI lists the sixty-eight magazines with the percentage
of monosyllables found in each. They are arranged from the
highest percentage of monosyllables, 75.8, in Ranch Romances,
to the lowest percentage, 61.5, in Review of Reviews. By apply-
ing the data shown in Table XXXVII to those given in Table
XLI, it is possible to define concretely what is meant by certain
measures of variability in the former table. For example, we
may say that the range of difficulty represented by general
magazines extends from Ranch Romances to Review of Reviews.
Half of the general magazines studied are easier than Liberty ,

HOW ADULT READING MATERIALS DIFFER 159

TABLE XLI

PERCENTAGE OF MONOSYLLABLES IN THE READING MATERIAL OF
SIXTY-EIGHT GENERAL MAGAZINES

Magazine

Mono-
syllables

Magazine

Mono-
syllables

Ranch Romances

75-8
75.2
75.0
74.2
74.1
73.6
73-5
73-2
72-3
7 2 -3

72.2
72.1
72.0
71.9
71.9
7*-7
71-4
71-4
70.9
70.8

70.6
70.5
70.5
70.5
70.4

70.3
70.1
70.1
70.0
70.0

70.0
69,6
69.5
69.4
69.0

Collier's

68.9
68.7
68.6
68.5
68.5
68.4
68.3
68.3
67.9
67.8

67-7
67.7
67.7
67.7
67.6
67.5
67.5
67.4

67.3
67.2

67.0
66.5
66.4
66.4
66.2
65.8

64.3
64.2
63.6
63.0

62.2
62.0
61.5

True Story

Blue Book

Dream World

Motion Picture Classic

All-Story

Golden Book

People's Popular Monthly

Woman's World

True Romances

Needlecraft

Short Stories

Capper's Farmer

All-Star Detective Stories.

Photoplay

Clues . .

Country Gentleman

Delineator

Holland's

Cosmopolitan

^lotion Picture

Aces

Picture Play

Pictorial Review

Science and Invention

Ace-High

Screen Book

Gentlewoman

National Geographic

Detective Story

Atlantic Monthly . . .

College Humor. . ...

True Detective IVIysteries

McCalTs

American Home

Red Book

Successful Farming

Good Housekeeping

S creenland

Ladies' Home Journal

Popular Mechanics

Farmer's Wife

New Movie

Field and Stream . . ,

Rotarian .... . ...

Tudffe

Venue

Holiday

National Farm Journal

Household ...

Nation's Business

American

Literary Digest . .

Woman's Home Companion
Better Homes and Gardens ....
Saturday TRv^pjmr Po^t

Popular Science

Pathfinder

House and Garden

World's Work

Radio News

Sky Riders

Time

Outdoor Life

Review of Reviews

Mtother's Home Life

Mean 69

.24.28

.37-2o

S.D 3

160 WHAT MAKES A BOOK READABLE

so far as syllabic word-length is concerned, and half are more
difficult than Mother's Home Life, since the median percentage
of monosyllables is located between these two magazines. One-
fourth of the magazines in this class contain more monosyl-
lables, on the average, than College Humor or McCalVs^ bound-
aries for the third quartile; whereas one-fourth contain a small-
er percentage of monosyllables than Atlantic Monthly y which
marks the first quartile.

In order to give a still more concrete notion of the degree of
variability indicated by the statistical measures in Table
XXXVII and at the same time illustrate differences among
individual magazines listed in Table XLI, we have shown graph-
ically in Figure n the percentage of monosyllables in the thou-
sand-word sampling of five magazines. These periodicals repre-
sent points of variability among the magazines studied. Ranch
Romances represents the highest point in the range of monosyl-
lables and is therefore considered the easiest magazine; College
Humor represents the third quartile; Mother's Home Life y the
median; Atlantic Monthly -, the first quartile; and Review of Re-
views, the lowest point in the range. The last-mentioned maga-
zine is presumably the most difficult from the point of view of
word-length. The percentage of monosyllables is shown for ten
hundred-word samplings, in ascending order from lowest to
highest. For example, in Ranch Romances, there are 64 per cent
of monosyllables in one hundred-word unit, 72, per cent in each
of two hundred-word units, 73 per cent in one unit, and so on,
to the highest percentage, 83, occurring in one hundred-word
unit. Percentages for other magazines shown in the figure may
be read in the same way. The figure indicates that there is
marked variability among magazines with respect to the per-
centage of monosyllables they contain.

A study of differences among specific materials naturally leads
to a study of likenesses. That is to say, if we set out to discover
how magazines differ among others of their class in word-length,
we shall end by discovering which ones approximately agree in
word-length. This appears desirable, for if knowledge of the

111

J ^H f \L

iil

8 a

Fio. n. Percentage of monosyllables in 1,000 words in the reading material of
fire selected magazines.

1 62 WHAT MAKES A BOOK READABLE

characteristic features of reading material is to function for any
practical purpose, it should enable one to recommend to a
reader not just this magazine, but this, or this, or this.

Examination of Table XLI reveals seventeen magazines
whose percentage of monosyllables varies not more than I per
cent in either direction from the mean. They are: American,
Better Homes and Gardens, Blue Book, Capper s Farmer, Collier's,
Golden Book, Liberty, Mother s Home Life, Motion Picture Clas-
sic, Needlecraft, Outdoor Life, Photoplay, Saturday Evening Post,
Sky Riders, World's Work, Woman's Home Companion, and
Woman's World. It appears from this list that the reader whose
interest leads him to World's Work need not anticipate encoun-
tering any longer words on the average than he may find in
Saturday Evening Post or Liberty, since all belong to the group
tentatively designated of "average" difficulty. If the first of
these three magazines is more difficult than the other two, then
it is clear that other elements than word-length must contribute
to the difficulty.

Further study of the table shows that the reader whose taste
leads him to All-Story, Dream World, Ranch Romances, and
True Story is reading 75 per cent monosyllables; while he who
reads the more serious content of Radio News, Review of Reviews,
and Time is reading material in which the percentage of mono-
syllables is lowest among general magazines. The first four are
obviously easiest of general magazines, and the last three, the
hardest, in so far as percentage of monosyllables may be taken
as a single index of difficulty.

Despite the fact that we have identified only a few magazines
as "easy," "average," or "difficult," we do not presume to have
exhausted the classification, for the reason that we have defined
categories arbitrarily. To be classified in any one of these cate-
gories a magazine must contain monosyllabic words within I per
cent of some particular statistical measure. We might quite as
well have said within 2 per cent or 3 per cent, except for the
reason that classifications are likely to be misleading if they are
not defined rather rigorously. The relative difficulty of each

HOW ADULT READING MATERIALS DIFFER 163

magazine can be defined above or below the mean, near to an
extreme or far from it,

DIFFERENCES IN SYLLABIC WORD-LENGTH IN
MAGAZINES OF DIFFERENT CONTENT

We now turn to consider the occurrence of monosyllables and
polysyllables in three different classes of magazines: magazines
of general fiction, of information, and of woman's interests. The
basis of these classifications is given by Ayers.

111

/a -O
O CJ

Monosyllables ----- Polysyllables

FIG. 11. Proportionate occurrence of monosyllables and polysyllables in maga-
zines of general fiction.

Figures 12, 13, and 14 give profiles of the three classes of mag-
azines with respect to the mean percentage of monosyllables and
polysyllables contained in them, when compared with the mean
percentage for all magazines. The base line of each figure repre-
sents the mean of all syllabic word-lengths. The word-length
for each magazine is expressed in terms of units of standard
deviation, 8 which are shown in the figure on an absolute scale.

8 The following formula was used in transmuting the means to units of standard
deviation: ^ Q , In which X indicates the mean of the magazine; Mx y the mean of
the distribution; and ?.D., the standard deviation. See Holzinger, op, cit. 9 pp. 118-22.

164 WHAT MAKES A BOOK READABLE

Lines extending above the mean indicate positive deviations;
lines extending below the mean, negative deviations. The un-
broken line represents monosyllables; the broken line, polysyl-
lables. Bisyllables and trisyllables are omitted from the figures
for the sake of clarity.

An analysis of the profiles shows the magazines which are
of average difficulty with respect to word-length and those

S 5 "R *

i titsji.

'rri IJi-HSs

,3 *4S * *** *** ^i "5p S SJ ^ *"^

Monosyllables Polysyllables

FIG. 13. Proportionate occurrence of monosyllables and polysyllables in maga-
zines of general information.

which are relatively easy or difficult. For example, Collier's ap-
proaches the mean for all magazines, and may be said to be of
"average" reading difficulty. Similar observations may be made
in regard to other magazines whose relation to the mean shows
relatively the same narrow deviation. Distinctly easy reading
in terms of word-length is shown for All-Story, Dream World,
Ranch Romances, Short Stories, and True Story. For these maga-
zines the deviation of monosyllables is from one to two sigmas

HOW ADULT READING MATERIALS DIFFER 165

above the mean, and the deviation of polysyllables approxi-
mately as far below the mean. For Atlantic Monthly,, Nation's
Business, and Review of Reviews, deviations are in the opposite
direction. Low negative deviations indicate a relatively small
percentage of monosyllables. High positive deviations indicate
a percentage of polysyllables conspicuously above the mean for
all magazines.

Monosyllables Polysyllables

FIG. 14. Proportionate occurrence of monosyllables and polysyllables in maga-
zines of woman's interests.

An inspection of these figures leads to fairly definite conclu-
sions with respect to monosyllabic and polysyllabic words found
in magazines of similar content. Figure 12 shows that magazines
of general fiction, with the exception of Atlantic Monthly and
Blue Book, have a generally high positive deviation from the
mean, when monosyllables are considered. This indicates that
they contain one-syllable words considerably above the average
word-length in all magazines. The figure also shows a generally
low negative deviation with respect to polysyl ? ables, indicating
that, with the few exceptions already noted, these magazines
contain a low percentage of words longer than three syllables.

In contrast, magazines of information, as shown in Figure 13,
contain a relatively low percentage of monosyllables and a high

166 WHAT MAKES A BOOK READABLE

percentage of polysyllables. Woman's magazines, as shown in
Figure 14, tend to deviate slightly, but with fair consistency, in
a positive direction for monosyllables and in a negative direction
for polysyllables. In general, the content in these magazines is
neither so easy as is the content in magazines of general fiction
nor so difficult as that in magazines of information.

The major purpose of these figures- is to present facts graphi-
cally. Yet they do more. They raise questions for which we
have no answer. For example: Who are the readers of All-Story,
Ranch Romances, True Story, Dream World, True Romances,
Short Stories, and other "easy" magazines? Do they read cheap
fiction because it offers escape from reality, as is commonly be-
lieved? Or do they read it because it is better adapted to their
level of reading ability than the higher quality found in Atlantic
Monthly or Golden Bookt Have they interest in the content of
Review of Reviews, Radio News, Nation's Business, or Popular
Science! Do they read these magazines as well as the easier
ones ? If not, would they read them were the vocabulary sim-
plified to the level of True Story and Ranch Romances ? But we
are not dealing with readers primarily. Our concern is chiefly
with reading materials. The most we can say in answer to such
questions is that no other magazine material approaches the
ease of easiest cheap fiction in so far as ease is measured by
length of word,

GENERAL MAGAZINES CLASSIFIED BY PERCENTAGE
OF DIFFERENT WORDS

Table XLII presents the sixty-eight magazines in order of
difficulty from Gentlewoman, with 44 per cent of different words,
to Time, with 55.9 per cent. Twenty-six magazines contain a
percentage of different words which varies not more than I per
cent in either direction from the mean, that is, between 49.5
and 51.5. They are Ace-High, All-Story, American, American
Home, Better Homes and Gardens, Clues, Cosmopolitan, Farmer's
Wife, Field and Stream, Holland's, Household, Liberty, National
Farm Journal, New Movie, Outdoor Life, Picture Play, Popular

HOW ADULT READING MATERIALS DIFFER 167

TABLE XLII
PERCENTAGE OF DIFFERENT WORDS IN SIXTY-EIGHT GENERAL MAGAZINES

Magazine

Percent-
age of
Different
Words

Magazine

Percent-
age of
Different
Words

Gentlewoman

44.0

44-1
46.6
46.6
46.8
47.0
47-4
47-6

47.8

47-9

48.2
48.4

48.4
4 8.6

48.7
49.0
49.0
49-4
49-5
49-5

49-5
49-7
49-7
49-7
49-7
49-8
49-9
50.0
50.0
50.0

50.2

50.3
50.4

50,4

Cosmopoli tan

50.6
50.8
51.0
51.0
51.1
51.1
51-3
51-3
51-4
51-5

51.6
51.6

5 J -7
51.8

5i-8
52.0
52.1
52.1

5 2 - 1
52.2

52.3
52.6

53-i
S3-*
53-5
53-6

53-7
54-i
54-1
54.2

54.3
54-5
55 -
55-9

Dream World

True Detective Mysteries

True Romances

Household

Capper's Farmer

Ace-High

Aces . ... ...

Short Stories

Detective Story

National Farm Journal . .

Ranch Romances

American .

Radio News

Sky Riders

Rotarian

Popular Mechanics

People's Popular Monthly

Farmer's Wife

Holiday

Judge

Needlecraft

Woman's World ...

Nation's Business

Review of Reviews

Delineator

Red Book

College Humor

Saturday Evening Post

True Story

Atlantic Monthly

All-Star Detective Stories

Collier's

Photoplay

Mother's Home Life . .

Woman's Home Companion
Better Homes and Gardens

World's Work

McCall's

Screen Book

Ladies' Home Journal . .

Successful Farming

Motion Picture Classic

Field and Stream

Golden Book

All-Story

Pictorial Review

Popular Science Monthly .

Blue Book

Science and Invention

Vogue

American Home

Motion Picture

Clues

Country Gentleman

New Movie .

House and Garden

Liberty

Pathfinder

Screenland

National Geographic

Picture Play

Good Housekeeping

Outdoor Life

Literary Digest

Holland's

Time

Mean 5

o.5.i 9
41 -i3

S.D 2

168 WHAT MAKES A BOOK READABLE

Mechanics, Popular Science Monthly, Science and Invention,
Screenland, Short Stories y Sky Riders, Successful Farming, True
Detective Mysteries, Woman's Home Companion, and World's
Work. So diversified is the content represented here that an
"average" magazine can be selected readily to suit a wide vari-
ety of tastes and interests.

Two magazines, Gentlewoman and Dream World, contain the
lowest percentage of different words and to that extent are
designated "easiest" among the magazines studied. Two maga-
zines at the upper end of the scale, namely, Time and Literary
Digest, are ranked of greatest difficulty with respect to percent-
age of different words.

Although there is a tendency for magazines of general infor-
mation to utilize a larger percentage of different words than
magazines of fiction, the tendency is not consistent enough in
itself to warrant the generalization that informational maga-
zines are more difficult than fiction. However, the former have
been found by the earlier analysis to contain a percentage of
monosyllables consistently below the mean. There is some evi-
dence, then, for the observation that the content of information-
al magazines is less simply expressed than is that of fiction
magazines.

GENERAL MAGAZINES CLASSIFIED BY AVERAGE
LENGTH OF SENTENCES

It is always difficult to make reliable distinctions based on
central tendencies, but it is especially complicating when the
basis of comparison is, for example, the median of 200 sentences.
If we say that the median length of sentences in Aces is 13 syl-
lables and in Review of Reviews 33 syllables, all we have shown
is that in each magazine half the sentences are longer and half
are shorter than the median length. How much longer or how
much shorter they are in their respective halves is not very
clear. A graphic presentation of the length of sentences sampled
from the two magazines shown in Figures 15 and 16 makes the
difference implied by the two medians far more intelligible.

Page

3
6

21
!2 4
27
30

33
36

39
42

45
4 8

5*
54

57
60

63
66
69

7 2
76
78
81
84

87
90

93
96

99
ioa
104

20 30 40 50 60 70 Syllables

FIG. 15, Analysis of the first 175 sentences in the aoo-sentence unit selected from
the reading material of Aces.

3 40 5

7 8o 90

no 100 Syllables

Flo. 1 6. Analysis of the first 175 sentences in the 2oo-sentence unit selected from
the reading material of Review of Reviews.

HOW ADULT READING MATERIALS DIFFER 171

But to attempt distinctions based on the number of syllables
in each of the 200 sentences in all the materials analyzed is ob-
viously impossible. We have, therefore, based our conclusions
regarding the relative difficulty of different magazines on the
median sentence-lengths shown in Table XLIII.

Seventeen magazines with a median sentence-length varying
not more than two syllables in either direction from the com-
posite median are: American, Blue Book, Capper's Farmer, Col-
lier s> Household, Judge, Liberty, McCall's, Mother's Home Life,
Motion Picture, New Movie, Photoplay, Picture Play, Saturday
Evening Post, S Greenland, Screen Book, and Woman's World.
All of the motion-picture magazines and several of the so-called
"home" magazines are included in this "average" group. Short-
est sentences as determined by their median lengths occur in
Aces, All-Story, All-Star Detective Story, Pictorial Review^ and
Detective Story. These are the easiest general magazines. The
hardest magazines, with respect to length of sentence, are
American Home, Atlantic Monthly, Radio News, Review of Re-
views, Rotarian, and Science and Invention.

That the easiest magazines are generally fiction, and the
hardest, magazines of information, is shown in Figures 17 and
18. As may reasonably be expected, magazines of general fiction
show less consistent deviation with respect to sentence-length
than magazines of information. The former group includes both
wood-pulp magazines, whose chief virtue is simplicity, and
quality magazines like Atlantic Monthly, whose chief defect is
complexity.

GENERAL MAGAZINES RANKED BY PERCENTAGE
OF SIMPLE SENTENCES

Table XLIV presents the facts concerning the percentage of
simple sentences in different magazines. The area of difficulty
defined by simple sentences extends from Sky Riders, with 62,
per cent, to Atlantic Monthly, with 27 per cent. These two maga-
zines represent quite distinct degrees of difficulty. The first con-
tains from 3 to 5 per cent more simple sentences than the next

172

WHAT MAKES A BOOK READABLE

TABLE XLIII

RANGE OP SYLLABIC SENTENCE-LENGTH AND MEDIAN SENTENCE-LENGTH
IN SIXTY-EIGHT GENERAL MAGAZINES

Magazines

Range

Me-
dian

Magazines

Range

Me-
dian

Aces

CC I

1*1.0

Photoplay

71-1

22.

All-Story

jj
60-1

I7.O

New Movie

72-1

22.

All-Star Detective Story

76-1

Picture Play

Q7-2

22. C

Pictorial Review

i\ji

IO1--I

Capper*s Farmer

81-4

2^.O

Detective Story

* w o

7O I

T/f O

Judge

QI-I

21 .0

Clues ....

/CTI

62-2

I<.O

Woman's World

9T~ a

23.0

Sky Riders

<A I

K.O

Successful Farming

98-4

24.0

Ranch. Romances

61-1

1< O

Farmer's Wife

1 1 1-2

ac.o

Gentlewoman

89-2

* J ' w
1C .C

Ladies' Home Journal. . . .

140-2

25.0

College Humor

1 2 1-2

16.0

Golden Book

1 7 1-1

2C.O

People's Popular Monthly .

7-1-1

16.0

Country Gentleman

14.1 I

2C.O

Dream World

72-1

17 O

Time

IOI 2

2C.C

Ace-High

I* *
64 I

*/ w
17.

National Geographic

01 I

s *
26.O

True Story. , ,

1 06-2

17.0

National Farm Journal . . .

IOI-3

26.O

Delineator.

IIC-2

17.

Better Homes and Gardens

84.-!

27.O

Cosmopolitan

IIC-2

18 o

Outdoor Life

I^A~2

iS.O

Red Book .

** j *
Q2-I

18.0

Popular Science

714

28.O

Miotion Picture Classic

87-1

18 c

Holland's

JO^2

28 o

True Romance

77-1

w j
i8.c

Holiday

J.V-/J

IOO;-!

28. <

Woman's Home Compan-

Vogue

IOT-2

2Q O

ion

QO-2

10.

*y -^

Field and Stream

8o-2

2Q <

Good Housekeeping

6o-I

10.

House and Garden

IOI -2

y ' j

1O. O

True Detective Mysteries

lOQ-I

IQ.O

Needlecraft

in i

7O <

Short Stories

77-1

IQ.O

Nation's Business

1 80-1

J w J
11 .0

Screen. Book

02 I

20.

Popular Mechanics

no 2

11 O

Motion Picture

IOO-I

20.0

Pathfinder

106-2

11 .0

Collier's

81-1

2O. O

World's Work

IOQ I

11 O

Household . , , ,

Q2-2

2O, C

Literary Digest

* 7 J.

no-2

11 .0

Blue Book

08-1

20. C

American Home.

82-1

12 O

Liberty

1022

2O C

Science and Invention

lA.11

rtn o

McCaU's

06 I

21 .O

1 T X

J-* - v

Atlantic Monthly

2O2-I

12.

Saturday Evening Post. . .

C2-I

21 .0

Rotarian

182-2

11 .0

Screenland

I2C-I

Radio News

lAl*-A

11 O

Mother's Home Life

<X 2

21 .0

Review of Reviews ....

121 1

JJ - u
11 . C

Amprtfflti

no n

rt r

oo * ,)

93-2

Composite median .

21 C

** O
4..Q

1
t

/'x

1 1 1 ^ I 1 18 I

FIG. 17. Syllabic length of sentences in magazines of general fiction

- v

FIG. 1 8. Syllabic length of sentences in magazines of general information

WHAT MAKES A BOOK READABLE

TABLE XLIV

PERCENTAGE OF SIMPLE SENTENCES IN THE READING MATERIAL
or SIXTY-EIGHT GENERAL MAGAZINES

MAGAZINE

TYPE

Simple

Sky Riders . .

62.0

American Home

4.C.C

Time

CQ.O

American

AC.C

People's Popular Monthly.

jy m ^
C7.O

House and Garden

4<.O

All-Star Detective Stories

*t
r6. C

Nation's Business

AA , C

Popular Mechanics

3 W - ^
C4..O

Collier's

4.4.. O

Ace-High

JT- 4V/

<4.O

College Humor

4?. ^

Clues ....

<4..O

Woman's World

4,*^ O

New Movie *

Blue Book

4^.0

Woman's Home Companion. . .

<2.{

Ladies' Home Journal

4^.O

Pictorial Review

All-Story

4.7.

Successful Farming

<2.O

Saturday Evening Post

4.2-. O

Motion Pictures. . . .

Science and Invention

4.2.

Aces. . .

<I <

Delineator

4.2.

Capper's Farmer

n .0

Gentlewoman

Short Stories.

<I ,O

Radio News . .

4.1 <

Needlecraft

j * v
to. c

Tudff e

*** 5
4.1 . <f

Pathfinder

<o.o

Mother's Home Life

4.1 .0

Detective Story Magazine

co.o

Red Book

4.1 .O

Better Homes and Gardens . .

AQ r

Cosmopolitan. . .

Al O

Country Gentleman

f T7' J

4,0.

Literary Digest

4.O. C

Screen Book

4,0.0

Holiday ,

4.O C

Popular Science Monthly

48.0

Good Housekeeping ,

4.O.O

Review of Reviews

4.7 <

Household

1Q C

National Geographic

*P/ O
4.7. <

True Detective Mysteries

oy 3

^Q C

Photoplay ,

T/ ' J

4.7.4

Holland's

jy o

*}Q <

Screenland

T 1 / * J

4.7. <

Golden Book

O7* J
*?o c

Farmer's Wife

TV ' J

4.7. C

Field and Stream . ...

*?6 ^

Outdoor Life ....

T 1 / O

4.7.X

True Story . . .

76 o

Rotarian

4/O

46. c

Picture Play

JW/.W

74.. C

National Farm Journal*

4.6. <

True Romances

JT" }
74. O

McCall's

46.0

World's Work

j<j,.w
74. O

Vogue

4.C.<

Ranch Romances. , *

77 O

Motion. Picture Classic

4.C.C

Dream World .

JJ * w

Liberty ,

T-J O

4.C.C

Atlantic Monthly

27 O

T"J J

*/ <w

Mean 4^

S.D 6

7^i .39

HOW ADULT READING MATERIALS DIFFER 175

easiest magazines, Time, People's Popular Monthly, and All Star
Detective Stories. And Atlantic Monthly contains about 6 per
cent less than Dream World and Ranch Romances, the next most
difficult magazines as defined by the occurrence of simple sen-
tences. Since the last two magazines have heretofore been
ranked "easy," the classification by simple sentences appears
contradictory. An examination of Table XLIV shows that other
fiction magazines also contain a relatively low percentage of
simple sentences. This, too, seems contrary to expectations.
One possible explanation lies in the fact that fiction, especially
of the cheaper sort, uses a large amount of conversation ex-
pressed in sentence-fragments, the meaning of which is implied
by the emotional tone of the conversation. For example, 9 per
cent of the sentences in Red Book are of this sort. In Ranch
Romances, sentence-fragments number 7.5 per cent; in Aces, 14
per cent.

In the average group with respect to simplicity of sentences
are ranked American, American Home, Collier's, House and Gar-
den, Liberty, McCall's, Motion Picture Classic, National Farm
Journal, Nation's Business, Rotarian, and Vogue.

GENERAL MAGAZINES IN COMPOSITE RANKING
OF DIFFICULTY

Having shown the relative difficulty of the magazines studied
by rankings on the basis of four elements of difficulty, we are
now in a position to show the composite ranking of these mag-
azines from all the facts available. Table XLV gives a com-
posite classification of magazines as "easy," "average," or "diffi-
cult." Although in general this classification represents the
trend of difficulty shown in earlier tables, some supplementary
data have been used not presented in this report. They were
obtained by analyzing doubtful magazines for occurrence of
different hard words, words not known to 90 per cent of sixth-
grade children, easy words, personal pronouns, and preposition-
al phrases.

Magazines classified in Table XLV are, therefore, those

I 7 6

WHAT MAKES A BOOK READABLE

which hold relatively the same position of difficulty by three or
more analyses. Magazines excluded from this classification
either fail to fall consistently within one of the three categories
or fall generally outside the areas designated "easy/* "average/*
or "difficult/* Their relative difficulty may be estimated rough-
ly from their rankings in Tables XLI-XLIV.

TABLE XLV

RELATIVE DIFFICULTY OF GENEILA.L MAGAZINES BY COMPOSITE

CLASSIFICATION ON THE BASIS OF FOUR

ELEMENTS OF DIFFICULTY

Easy

Aces
All-Story
Dream World
Clues

Gentlewoman
True Romances

Average

American

American Home

Better Homes and Gardens

Collier's

McCail's

Mother's Home Life

Capper's Farmer

Sky Riders

Picture Play

New Movie

Liberty

Motion Picture Classic

Cosmopolitan

Saturday Evening Post

Outdoor Life

House and Garden

World's Work

Difficult

Atlantic Monthly
Literary Digest
Pathfinder
Radio News
Review of Reviews

From the facts pertaining to the occurrence of significant ele-
ments of difficulty in general magazines, the following conclu-
sions seem warranted:

1. It is evident that definite strata of difficulty exist among
general magazines when difficulty is measured by the presence
or absence of certain structural elements.

2. There is marked agreement among rankings by individual
elements for those magazines defined as "easy/* "average," or
"difficult." Less agreement exists among magazines at inter-
mediary positions.

3. If a magazine consistently ranks of the same relative diffi-

HOW ADULT READING MATERIALS DIFFER 177

culty by three or more single elements, the use of a larger num-
ber of elements is usually unnecessary.

4. It seems significant to note that magazines which rank
"average" are the most popular magazines. They represent the
widest circulation among all magazines. One-third have a circu-
lation above 2,000,000; one-half, above 1,000,000; and two-
thirds, above 200,000. Several inferences may be drawn. One is
that their content is what most magazine readers want to read
about. Another is that the average difficulty of their content
is suited to the reading ability of most magazine readers. A
third is that editors of these "average" magazines definitely aim
to reach a large number of readers by means of interesting con-
tent, easy expression, popular price, and other factors. For ex-
ample, Liberty and American rank "average" by every measure
applied in this section. They are typical average magazines,
then, when difficulty is measured by structural elements sig-
nificantly related to difficulty.

5. Circulation figures for magazines designated "easy" show
that all are purchased and presumably read by more than
200,000 persons, and one by more than 1,000,000. We can only
conjecture as to whether readers of these magazines actually
prefer low-grade, all-fiction content; whether their ability to
read is best served by these magazines; or whether, having
found a certain pleasure in effortless reading, they are too in-
dolent to learn to like anything better.

6. Magazines rated most difficult tend to be in least demand
among adult readers. Less than one-half have circulation figures
above 200,000.

7. Writers on politics, world-news, and informational ma-
terial in general tend to use longer words and sentences than do
writers of fiction. With respect to percentage of different words,
they seem to follow no general practice. There is evidence that
although the thought-content expressed by these writers may
be harder than fiction, words used in expressing it may or may
not be greatly diversified.

8. Writers who make their appeal to the primary and ele-
mental interests of their readers tend to utilize the simplest

178 WHAT MAKES A BOOK READABLE

structural elements and, consequently, to produce materials
easiest to read.

HOW DO NEWSPAPERS VARY WITH RESPECT TO
ELEMENTS OF DIFFICULTY?

This question is answered by Tables XLVI-XLIX. In each,
the fifteen newspapers are listed in order of difficulty, as deter-
mined by the occurrence of a specific element of difficulty. For

TABLE XLVI

PERCENTAGE OP MONOSYLLABLES IN THE READING MATERIAL
OF FIFTEEN NEWSPAPERS

-_ Monosyl-

Newspaper lables

Chicago Daily Times .......................... 70. 6

Capper's Weekly .............................. 68.9

Riceville Recorder ............................ 67. 5

Chicago Tribune .............................. 65 .4

New York Evening Graphic .................... 64. 9

New York News .............................. 64. 8

Boston Post .................................. 64 . 4

Edmore Herald-News .......................... 64 . 3

Christian Science Monitor ...................... 62 . 7

Los Angeles Times ............................. 62.3

Chicago American ............................. 62.2

New York Journal ............................. 62 . o

Philadelphia Evening Bulletin ................... 62.0

Jefferson Banner .............................. 61.7

Olney Daily Mail .............................. 61 .3

Mean .............................. 64:3 ^47

S.D ................................

example, in Table XLVI the Chicago Daily Times is easiest and
the Olney Daily Mail, hardest, with respect to monosyllables.
In Table XL VII the Riceville Recorder, with 46.2 per cent of
different words, ranks easiest; and the Los Angeles Times, with
53.2 per cent, ranks hardest. Data presented in these tables
support the facts revealed by Tables XXXVII-XL to the effect
that newspapers as a class tend to agree fairly closely in their
use of elements of expression included in this study. For this
reason we have made no attempt to classify the fifteen news-
papers according to relative difficulty.

HOW ADULT READING MATERIALS DIFFER 179

In so far as the data warrant, we may define the relative diffi-
culty of newspapers as a class in terms of specific publications.
For example, the area of difficulty represented by the fifteen
newspapers studied extends from the level of Capper s Weekly
and the Chicago Daily Times to that of the Los Angeles Times
and Christian Science Monitor. Between these two extremes,
which are after all not conspicuously different, lies an interme-
diate area. This area is generally characterized by village week-

TABLE XLVII

PERCENTAGE or DIFFERENT WORDS IN FIFTEEN NEWSPAPERS

Percentage of

Newspaper Different

Words

Riceville Recorder 46 . 2

Capper's Weekly 48 .5

Edmore Herald-News 49 . 5

Chicago Tribune 50.3

Philadelphia Evening Bulletin 50.3

Chicago Daily Times 50.6

New York News 51 .o

New York Evening Graphic 51 .3

Boston Post 51 .6

Chicago American 51 .7

Jefferson Banner 51.9

Christian Science Monitor 52 . a,

Olney Daily Mail 52.5

New York Journal 53 . 1

Los Angeles Times 53 . 2

Mean 50.9 .31

S.D i .77 . 22

ly papers, such as Edmore Herald-News and Jefferson Banner,
the difficulty of which tends to merge on the one side with that
of graphics and on the other with the usual large daily news-
paper.

OCCURRENCE OF SIGNIFICANT ELEMENTS OF DIFFICULTY
IN DIFFERENT CLASSES OF NEWSPAPERS

Interesting differences are revealed by comparing the occur-
rence of elements of difficulty in newspapers published pri-

i8o

WHAT MAKES A BOOK READABLE

marily to serve a particular class of readers. Figure 19 shows
these differences in percentage of monosyllables and polysyl-
lables. In large daily newspapers, excluding graphics, monosyl-
lables tend to deviate negatively from the mean; and polysyl-
lables, positively. In graphics, the direction of the deviations is
reversed. In other words, the reader who extracts his news in

TABLE XLVIII

RANGE OF SYLLABIC SENTENCE-LENGTH AND MEDIAN SENTENCE-
LENGTH IN FIFTEEN NEWSPAPERS

Newspaper

Syllabic
Range

Median

Olney Daily Mail

82-2

21. <

Capper's Weekly

122-4

22.

New York Kvening Graphic

26 c

New York News

QO-I

27.0

Chicago Daily Times

20 O

Los Angeles Times

112-4

11 O

Chicago American

8 1-7

11 O

Riceville Recorder.

110-6

33 o

Jefferson Banner

126-4

14 O

Boston Post

II4-1

34 *^

New York Journal

IOI 4

35 o

Philadelphia Evening Bulletin

16 o

o6-{

Christian Science Monitor. ...

126

Kdmore Herald-News

*\J 2
I4O I

4 -5

Composite median .

literal form from the usual daily newspaper reads a larger per-
centage of long words than does the person who reads the pic-
ture newspaper with its minimum of print. Does the latter
reader prefer the picture newspaper because pictures provide
easy reading or because short words make reading easy? Or
is there another factor than ease that accounts for the pref-
erence? Again, the question suggests the kind of information
needed to explain why people read what they do.
The easiest paper illustrated in the figure is the Chicago Daily

HOW ADULT READING MATERIALS DIFFER 181

Times; the hardest, the Christian Science Monitor. Inasmuch
as village and rural papers show no consistent tendency toward
simplicity or complexity as measured by word-length, they are
not given in the diagram.

Despite the close agreement among newspapers with respect
to percentage of different words shown in Table XL VII, there

TABLE XLIX

PERCENTAGE OF SIMPLE SENTENCES IN THE READING
MATERIAL OF FIFTEEN NEWSPAPERS

TYPE OF SENTENCE

NEWSPAPER STRUCTURE

Simple

Los Angeles Times 50 .o

Olney Daily Mail 50 .o

Capper's Weekly 49 .o

New York Evening Graphic 47 ,o

Jefferson Banner 44 .o

New York News 44.0

Chicago Daily Times 44 .o

Riceville Recorder 42 .o

Philadelphia Evening Bulletin 42.0

Chicago American 40 . o

Chicago Tribune 38 .o

Christian Science Monitor 37 .o

Boston Post 37.0

New York Journal 30.0

Edmore Herald-News 28 .o

Mean 4i.46i.i2

S.D 6.42! .79

is a tendency for the vocabulary of village and rural papers to
be slightly less diversified than that of large daily papers. Al-
though the difference in range of different words is on the aver-
age no more than 7 per cent, it is enough to show the influence
of cosmopolitan news on difficulty.

When lengths of sentences are compared (Fig. 20), it may be
noted that large daily newspapers, exclusive of graphics, on the
average contain longer sentences than the mean sentence-length
for all newspapers. Graphics, on the other hand, employ shorter

/f\

* 1 -- .

-*-**

--''I

>v"X

->s^__

.BorfoH POJ/
Chicago America

cfc'ogg* n//^ T

^ Christian Scienc

New York Jourr

LOJ Angeles Tin

j Philadelphia tt

e
^>

New York

I
I

FIG. 19.- Proportionate occurrence of monosyllables and polysyllables in daily
newspapers,

I
3

I I

i l

Fio. ao. Syllabic length of sentences in daily newspapers

HOW ADULT READING MATERIALS DIFFER 183

sentences than the common mean. The longest sentences occur
in Christian Science Monitor, which also contains the lowest
percentage of monosyllables and a diversity of vocabulary above
the median. No generalization can be made with respect to the
length of sentences that characterize rural and village papers.
Two papers in this class, however. Capper s Weekly and Olney
Daily Mail, use the shortest sentences among the newspapers
studied. The facts from which Figure 20 was derived seem to
suggest that rural and village publications employ a simplicity of
expression, as measured by sentence-length, more nearly like
graphics than other daily newspapers.

Distinctions between different types of newspapers with re-
spect to sentence-form may be obtained from Table XLIX.
The facts presented suggest that picture newspapers which con-
tain a percentage of simple sentences above the common mean
are probably easier than other metropolitan papers containing
a percentage of simple sentences below the mean. Rural and
village publications do not vary markedly in either direction
from common practice.

WHAT DIFFERENCES EXIST AMONG DEPARTMENTS
OF NEWSPAPERS?

Inasmuch as newspaper material on the whole tends toward
a consistent use of different structural elements, question may
be raised as to whether separate departments of newspapers
agree as closely* We, therefore, have analyzed word-lengths in
each of nine departments chosen for study with the following
results: (i) More agreement in the use of monosyllables exists
in "Local News" than in any other department. (2) Least agree-
ment is found in "Washington News." (3) The range of poly-
syllables is widest in "Editorials" and narrowest in "Local
News." (4) The highest percentage of polysyllables occurs in
"Washington News" and the lowest in "Local News." (5) Large
daily papers tend to use a lower percentage of monosyllables in
their "Local News" than do village papers. (6) Although indi-
vidual newspapers vary the proportionate word-lengths in dif-

184

WHAT MAKES A BOOK READABLE

ferent departments, there is no evidence to indicate that they
follow any definite practice except in "Local News."

DIFFERENCES AMONG TWENTY-NINE BOOKS WITH
RESPECT TO LENGTH OF WORD

Table L brings out the differences in syllabic word-length
found in an analysis of twenty-nine books. It will be remem-

TABLE L

PERCENTAGE OF MONOSYLLABLES IN THE READING
MATERIAL OF TWENTY-NINE BOOKS

Book

Percent-
age of
Mono-
syllables

Book

Percent-
age of
Mono-
syllables

Robinson Crusoe (adapted) ....
Carl and Anna

83.3
81 2

Reading and Living, Book III. .
Cimarron

72.9

72 6

Huckleberry Finn

70.7

Moby Dick (original)

72.

All Quiet on the Western Front

/?/
77 6

Silas Marner (original)

72 O

Reading and Living, Book I .

77. C

Poe's Tales

71 . 1

Robinson Crusoe (original)

76 Q

Scarlet Letter, The

6q 7

Bible

7C.Q

Byron

uy./
DO 1

Reading and Living, Book II . , ,

7<-Q

Preface to Morals, A

v y -J

68 q

Silas Marner (adapted)

7^.8

Henry the Eighth . ...

68 6

Moby Dick (adapted)

/ j w
7<.A

Last of the Mohicans, The, . .

68 c

Peder Victorious ...

7C.I

Art of Living The

68 4

Roper's Row

7{,O

Elizabeth and Essex

66 2

Angel Pavement ,

74.. 7

Tragic Era, The

61 .0

7- j *7

White Oaks of Jalna .

74-7

73 2

Mean 7*3

A 4- ff

Story of San Michele, The

7T.O

S.D A

T - 1 - bb

A2"f~ TO

4-* -t. -oy

beted that these books include popular novels, popular general
books, standard fiction, and simplified classics.

The area of difficulty shown in the table extends from the
adapted Robinson Crusoe, with 83.3 per cent of monosyllables,
to The Tragic Era, with 61.9 per cent. Three areas of difficulty
are rather sharply defined. In the "average" group are Angel
Pavement, Cimarron, Peder Victorious, Roper's Row, Scarlet
Sister Mary, White Oaks of Jalna, and Reading and Living,
Book III. What this statement seems to imply is that, with the

HOW ADULT READING MATERIALS DIFFER 185

exception of one novel, All Quiet on the Western Front., which is
apparently easier than those just listed, all popular novels
studied contain an average proportion of short words about like
that in a randomly selected reading text for sixth grade.

Books which contain the lowest percentage of monosyllables
belong to the class designated "popular general books," They
are: The Art of Thinking, A Preface to Morals, Elizabeth and

N O BS N ^ ^ %

Monosyllables Polysyllables

FIG. 11 . Proportionate occurrence of monosyllables and polysyllables in popular
novels.

Essex, and The Tragic Era. At the opposite extreme of the list
are Robinson Crusoe (adapted), Reading and Living, Book I,
Carl and Anna, Huckleberry Finn, and All Quiet on the Western
Front. These are the easiest books, when word-length is taken
as a single index of difficulty.

Figures 21 and 1*2 support the facts already presented, that
popular novels are about of average difficulty and popular gen-
eral books are most difficult among the books studied. The pro-
files of popular fiction show relatively little deviation from the
mean for monosyllables and polysyllables. Popular general
books, on the other hand, commonly show a negative deviation
for monosyllables and a positive deviation for polysyllables.

i86

WHAT MAKES A BOOK READABLE

BOOKS CLASSIFIED ACCORDING TO VOCABULARY DIVERSITY

The conclusion to be drawn from an analysis of extent of vo-
cabulary, Table LI, is that more agreement than variation per-
sists among the books studied, except in those whose writers
have consciously attempted to reduce the size of vocabulary for

* *

^ 4 j

Monosyllables Polysyllables

FIG. 22. Proportionate occurrence of monosyllables and polysyllables in popular
general books.

the sake of simplicity. For example, Carl and Anna, with basic
English vocabulary intended "to do the work of the 20,000 com-
mon words usually required for the same purpose," contains
25.5 per cent of different words. This is against 42.5 per cent,
the mean for all books studied, and 50.6 per cent for The Tragic
Era. Adapted versions of Robinson Crusoe and Silas Marner
rank close to Carl and Anna.
Few books vary in any considerable degree from the mean for

HOW ADULT READING MATERIALS DIFFER 187

all books studied. The most difficult has a vocabulary about as
diversified as a sixth-grade reader, Reading and Living, Book

TABLE LI

VOCABULARY DIVERSITY IN TWENTY-NINE BOOKS

Percentage of

Book Different

Words

Carl and Anna 25 . 5

Robinson Crusoe (adapted) 29 .6

Silas Marner (adapted) 35 .4

Bible 36.1

Robinson Crusoe (original) 37 . 6

Huckleberry Finn 37 .7

Preface to Morals, A 39 . 2

Moby Dick (adapted) 39.2

Reading and Living, Book I 41.7

Angel Pavement 41.7

All Quiet on the Western Front 41 . 8

Roper's Row 42. i

Peder Victorious 42 . 4

Scarlet Letter, The 43.7

Art of Thinking, The 44.4

Silas Marner (original) 44. 8

Scarlet Sister Mary 44. 8

Moby Dick (original) 45. i

Elizabeth and Essex 45 .4

Poe's Tales 45.9

Cimarron 45 . 9

White Oaks of Jalna 46.3

Last of the Mohicans, The 46. 8

Story of San Michele, The 46.9

Byron 47-3

Henry the Eighth 47.9

Reading and Living, Book II 48 . i

Reading and Living, Book III 48 .3

Tragic Era, The 50.6

Mean 42. 5 . 69

S.D 5-58 -49

III. In the "average" group belong The Art of Thinking, Cimar-
ron, Elizabeth and Essex, Moby Dick (original) , Peder Victorious,

i88 WHAT MAKES A BOOK READABLE

Poe's Tales, The Scarlet Letter, Scarlet Sister Mary, Silas Marner,
and White Oaks of Jalna. Again, most of these books are popu-
lar novels, the "averageness" of which is shown by their slight
variation from the mean for all books.

RELATIVE DIFFICULTY OF BOOKS DETERMINED BY
AVERAGE LENGTH OF SENTENCES

In the very scattered distribution of books based on median
length of sentences (Table LII), the original Robinson Crusoe
ranks highest with a median sentence-length of 87 syllables. A
Preface to Morals and the original Silas Marner rank next with
median sentence-lengths of 37 and 36.5 syllables, respectively. It
seems reasonable to assume that the extraordinary length of sen-
tences found in Robinson Crusoe does not imply the same sort of
complexity that characterizes A Preface to Morals. It rather il-
lustrates the simple run-on, unpunctuated thought-units typical
of the early tale. Such sentences are in reality simple, if the
reader supplies the internal punctuation necessary to distinguish
single thought-units.

Easiest books, that is, those containing the lowest average
length of sentence, are Roper's Row, Silas Marner (adapted),
and Angel Pavement. Popular fiction generally ranks below the
median length for all books, as exemplified by Reading and Liv-
ing, Book III, and Henry the Eighth. If Peder Victorious, Scarlet
Sister Mary, Cimarron, All Quiet on the Western Front, White
Oaks of Jalna, and other easy books are representative of their
class, then modern novelists may be said generally to employ
short sentences. Writers of popular non-fiction, on the other
hand, tend to use sentences somewhat longer than the median
length for all books studied.

BOOKS CLASSIFIED ACCORDING TO SIMPLICITY
OF SENTENCE FORM

When facts are considered relative to the structure of sentences
(Table LIII), Cimarron is found to rank highest in percentage
of simple sentences and the original Robinson Crusoe, lowest.

HOW ADULT READING MATERIALS DIFFER 189

The former, a modern story of adventure in the pioneer West,
is a swiftly moving, cinematic panorama of events; whereas the

TABLE LII

RANGE OF SYLLABIC SENTENCE-LENGTH AND MEDIAN SENTENCE-
LENGTH IN TWENTY-NINE BOOKS

Book

Syllabic
Range

Median

Robinson Crusoe (original)

2Q7 1 7

87 o

Preface to Morals, A ....

IC7 <

07 o

Silas Marner (original)

A iJ i

I D7 2

J/- u
06 c

Last of the Mohicans, The

176 7

1A. C

Art of Thinking, The

186- 7

74 O

Tragic Era The

IK A

77 O

Moby Dick (original) . . . .

J -*3 *r

1 86- i

72 C

Scarlet Letter, The

177 2

71 O

Story of San Michele, The

78- 2

7O. O

Poe's Tales

III i

2Q O

Bible

87- 2

28. c

Byron

Q8- 4

28.O

Elizabeth and Essex

118- 2

28.O

Henry the Eighth

17 C- 2

24.0

Reading and Living, Book III ....

04 2

24. O

Reading and Living, Book; II

66- 4

2O. O

Reading and Living, Book I

84- 7

IQ o

Huckleberry Finn

07 i

IQ O

Peder Victorious

86- 7

18.0

Scarlet Sister Mary '

81- i

18.0

Robinson Crusoe (adapted)

5<- <

18.0

Moby Dick (adapted)

85- i

17.0

Cimarron

74. 2

16.0

All Quiet on the Western Front

IIQ- 2

IC.O

White Oaks of Jalna

67- I

ic,o

Carl and Anna

8l- 2

K.o

Angel Pavement

143- 2

H.c

Silas Marner (adapted)

36- 7

I4.O

Roper's Row

66- i

12.

Composite median 24.0

Q 7-7

latter is a leisurely, autobiographical tale of two centuries ago.
Further analysis of these books shows that the former contains
not only a high percentage of simple sentences but more than

190 WHAT MAKES A BOOK READABLE

8 per cent of fragmentary sentences. In Robinson Crusoe^ on
the other hand, the latter kind of sentence is wholly lacking.

TABLE LIII

PERCENTAGE OF SIMPLE SENTENCES IN THE READING MATERIAL
or TWENTY-NINE BOOKS

Percentage of

Book Simple

Sentences

Cimarron 55 . 5

Carl and Anna 52.5

Moby Dick (adapted) 51.0

Roper's Row 51.0

All Quiet on the Western Front 46 . 5

Silas Marner (adapted) 46 .o

Byron 43-5

Reading and Living, Book II 43 . 5

Reading and Living, Book 1 40. 5

Henry the Eighth 40. o

Tragic Era, The 40.0

Art of Thinking, The 39.0

White Oaks of Jalna 39'

Scarlet Sister Mary 37. 5

Reading and Living, Book III 37 .o

Story of San Michele, The 35.5

Huckleberry Finn 34. 5

Angel Pavement 33 . 5

Peder Victorious 33 .o

Robinson Crusoe (adapted) 32 .o

Poe's Tales 30.5

Scarlet Letter, The 29.0

Elizabeth and Essex 28,5

Moby Dick (original) 27 ,o

Preface to Morals, A 23 . 5

Last of the Mohicans, The 20.0

Holy Bible 18.0

Silas Marner (original) 14.0

Robinson Crusoe (original) 4.0

Mean 35, 4 1.46

S.D ii-7i.o3

Over 50 per cent are long, loosely constructed, compound-com-
plex sentences.

HOW ADULT READING MATERIALS DIFFER 191

The average percentage of simple sentences for all twenty-
nine books is approximately the same as for Reading and Living,
Book III. It is apparent, therefore, that "average" books, as
defined by sentence-structure, namely, The Story of San Michele,
Scarlet Sister Mary, Angel Pavement, Peder Victorious, and
White Oaks of Jalna, tend to utilize a complexity of expression
about equivalent to that of a sixth-grade text in reading. As in
earlier classifications, popular novels, with some exceptions,
generally comprise the "average" group. This indicates that
their writers avoid extremes of simplicity or complexity with
respect to certain significant elements of expression. Inasmuch
as writers of popular informational books use a percentage of
simple sentences but slightly below the mean, we should prob-
ably not accept too readily the notion that simple sentences are
best adapted to simple ideas. Obviously, difficult ideas may be
expressed in fairly simple sentences, as popular writers of non-
fiction exemplify. Conversely, relatively simple ideas may be
expressed in such involved sentences as we find in The Last of the
Mohicans, the Bible, the original Moby Dick, Silas Marner, and
Robinson Crusoe.

COMPOSITE CLASSIFICATION OF BOOKS FOR
RELATIVE DIFFICULTY

What is the agreement represented by all these facts pertain-
ing to the relative difficulty of the books studied when difficulty
is defined by isolated elements? What contribution can be made
by analyzing the books for other elements of difficulty?

Table LIV brings together all the available evidence bearing
on these two questions. It shows the relative difficulty of books
as revealed by the analysis of the four elements just presented
and by the additional analysis of four other elements. The first
four are, again, percentage of monosyllables, percentage of differ-
ent words, syllabic sentence-length, and percentage of simple
sentences. The other four are number of different hard words,
percentage of easy words, number of personal pronouns, and

192

WHAT MAKES A BOOK READABLE

number of prepositional phrases. Details of the supplementary
analysis for the occurrence of the last four elements have not
been included in this report.

TABLE LIV

BOOKS DEFINED AS RELATIVELY "EASY," "AVERAGE," OR "DIFFI-
CULT" BY THE OCCURRENCE OF EIGHT ELEMENTS
OF DIFFICULTY

Easy

Average

Difficult

Robinson Crusoe (adapted)
Silas Marner (adapted)
Carl and Anna
Reading and Living, Book I

Angel Pavement

Bible

Byron

Cimarron

Moby Dick (original)

Scarlet Sister Mary

All Quiet on the Western

Front

Roper's Row
Peder Victorious
White Oaks of Jalna
Reading and Living,

Book III

Art of Thinking, The
Elizabeth and Essex
Last of Mohicans, The
Preface to Morals, A
Tragic Era, The

SUMMARY

The aim of this chapter was to analyze adult magazines, news-
papers, and books for the purpose of determining differences
among reading materials of a class with respect to certain ele-
ments which indicate their relative ease or difficulty. The anal-
ysis was directed toward a study of four significant elements:
percentage of monosyllables, percentage of different words, syl-
labic length of sentences, and percentage of simple sentences.

In interpreting results of the analysis, we have designated
certain materials as "average," "easy," or "difficult/' A maga-
zine of average difficulty, for example, is one in which the oc-
currence of a particular element approximates the mean occur-
rence of that element for the sixty-eight magazines studied.
Deviations from the mean in the direction of simplicity or com-
plexity mark other materials as easy or difficult respectively.

HOW ADULT READING MATERIALS DIFFER 193

Inasmuch as this designation of materials as "easy" or "diffi-
cult" is purely relative, it is clear that a particular magazine,
for example, may be ranked as "easy" when compared with
certain other magazines, yet be ranked "difficult" when com-
pared with still others. Hence, we cannot say that House and
Garden is a difficult magazine, except in comparison with other
of the sixty-eight general magazines studied. Neither can we
say that the Los Angeles Times is a difficult newspaper nor Eliza-
beth and Essex a difficult book, save as we compare each with
other materials of its class. Unquestionably, the classification
of House and Garden as difficult is more reliable than a similar
classification of the Los Angeles Times and Elizabeth and Essex,
for the reason that general magazines received a far more com-
prehensive sampling than newspapers and books.

The facts presented in this chapter, then, do not define diffi-
culty of reading materials precisely. They show, rather, how
differences in the occurrence of certain elements in several ma-
terials of a class warrant the conclusion that one book is easier
or more difficult than others with which it is compared. Further-
more, they show that differences which persist among different
reading materials with respect to significant elements are so ap-
parent as to suggest the need of a more precise technique of
defining the degree of ease or difficulty of specific books, maga-
zines, and newspapers. Such a technique involving a more ex-
tensive sampling and the use of a larger number of elements
has been presented on pages 134-138 of the previous chapter.
The task of applying that technique to particular reading ma-
terials is described in the next chapter.

CHAPTER VI

WHAT IS THE DIFFICULTY OF ADULT
READING MATERIALS?

PRECISELY how difficult are adult reading materials of
a general nature, when difficulty is expressed in terms of a
numerical index? If they differ widely (and we have
shown that they do) with respect to certain inherent elements
which mark them as relatively easy or difficult, what are the
limits of difficulty represented by them ? At what point along
the scale of difficulty can we say that a book is "very easy" and
not just "easy/' or 'Very difficult" rather than "difficult?" And
at what point is a book to be considered of "average" difficulty?
Having defined areas of difficulty numerically, how are we to
interpret them? That is to say, if the index of difficulty of New
Russia's Primer^ for example, shows that it is "easy," what does
this mean? Is an easy book about equal in difficulty to a book
for second grade, for fourth grade, or for eighth grade?

The foregoing questions arise as one surveys adult reading
materials and compares the occurrence of elements of difficulty
contained therein, as we have done in chapter v. The impor-
tance of these questions cannot be overemphasized. If we are
to utilize most profitably the materials now available, it is evi-
dent that the difficulty of each must be defined precisely. This
cannot be accomplished merely by showing, as we have done in
chapter v, that a certain book contains as many elements of
difficulty as this one or that one, but by determining its index
of difficulty and thereby defining the area of difficulty to which
it belongs.

In order to answer these questions, two sets of facts are need-
ed. First, facts should be discovered that can be used in defining
both the scale of difficulty represented by a wide sampling of
adult materials and the various areas of difficulty into which the

194

DIFFICULTY OF ADULT READING MATERIALS 195

entire scale can be divided. Second, facts are needed that can
be used in interpreting each area of difficulty in practical terms.
Obtaining these different kinds of information involves the se-
lection of a wide sampling of reading material, the calculation
of an index of difficulty for each of the materials selected, and
the distribution of the indexes in a manner that will make them
meaningful and useful. The present chapter shows how these
procedures have been carried out and the facts thus obtained
used in answering the questions presented at its beginning.

LIMITATION OF ADULT MATERIALS TO GENERAL BOOKS

In this part of the study which defines areas of difficulty for
adult reading materials, we have centered our attention on
books to the exclusion of magazines and newspapers. The rea-
sons for doing so are given in chapter v. An examination of
Tables XXXVII-XL shows that the range of occurrence of the
four elements studied is wider for adult books than for adult
magazines and newspapers. Presumably, a similar variation in
range persists with respect to other significant elements. If,
then, we define areas of difficulty for a wide sampling of adult
books, we may expect to include a range as wide as that repre-
sented by adult materials in general, that is, by books, maga-
zines, and newspapers.

ADULT BOOKS STUDIED

In selecting books for study, we have endeavored to obtain
a sampling that represents a range of difficulty for adults of
limited reading ability from very easy to very difficult. Such a
range extends from an area of difficulty exemplified by the sim-
plified Ro&inson Crusoe> through an area characterized by the
novels of Zane Grey, Harold Bell Wright, and other writers of
popular fiction, to an area represented by BoswelFs Life of
Johnson. The books studied were suggested by the following
sources:

I. The largest number were taken from a compilation pre-
pared to meet the demand for a list of simply written, informa-

196 WHAT MAKES A BOOK READABLE

tive, readable books suitable for use in connection with current
educational activities among adults. 1 The list includes books
of fiction, biography, history, and travel; books about cooking,
gardening, dress making, and other practical arts; and books
on music, drawing, painting, writing, the theater, and other
subjects bearing on the cultural aspects of living.

a. A considerable number were suggested by forty readers'
advisers in public libraries in various parts of the country. They
kept a month's record of the reading carried on by adult patrons
who came for guidance and evaluated the readability of the
books they read by the technique described in chapter ii.

3. Data compiled by Waples and Tyler furnished the names
of cheap, simply written novels which represent the undirected
reading choices of adults of low reading abilities and relatively
untutored tastes.

4. The findings obtained by Chancellor from an inquiry con-
cerning available reading materials for native-born illiterates
and near-illiterates indicate the books of greatest usefulness for
adult beginners in reading. These books were also included in
the list studied.

Many books were found to be common to two or more lists.
Others appear on but one list. Several had been analyzed and
ranked for relative difficulty after the manner described in chap-
ter v. No book was omitted from the study, however, if it could
be located in public or private libraries. In all, 350 suggested
books were available. Their indexes of difficulty were predicted
and used in defining areas of difficulty for adult reading matter in
general. The names of these books, together with author, pub-
lisher, and date of publication, are listed in Table LXXXV in
Appendix R

PREDICTING DIFFICULTY IN TERMS OF AN
AVERAGE READING SCORE

The method used to determine difficulty in terms of a numeri-
cal index, or an average reading score, has been described on

1 Doris Hoit, Books of General Interest for Today's Readers. Washington, D.C., Amer-
ican Library Association and the American Association for Adult Education, 1934.
Pp. 60.

DIFFICULTY OF ADULT READING MATERIALS 197

pages 134-135 of chapter iv. This method enables us to com-
bine several elements of difficulty into a single value, jf x> in
such a way as to give the best possible estimate of X*. Here X* is
the average reading score that adults of limited reading ability
might be expected to make were they tested for comprehension
of a given book. In other words, it is the index of difficulty
which we are trying to determine.

In chapter iv we have shown how eight significant elements
may be combined into a variety of smaller groupings for the
purpose of predicting difficulty. Any one of the combinations
presented gives a fairly reliable prediction of difficulty. For pur-
poses of this part of the study, we have combined five elements
number of different hard words, number of personal pro-
nouns, average sentence-length in words, percentage of differ-
ent words, and number of prepositional phrases into a single
instrument of prediction, as shown on page 138, The final index
is a true measure of difficulty only to the extent that the five
elements represent total difficulty.

The procedure followed in using these elements for predicting
the index of difficulty for a particular book is, first, to select a
passage of one hundred words from approximately each chap-
ter; second, to analyze the passage for the occurrence of the five
elements of difficulty; third, to calculate the average occurrence
of each element for the book as a whole; fourth, to substitute
the obtained values in respective order in the following equa-
tion:

Xi = . 01029^4- . 00901 2^T 5 . 02094^6 . 03313-^7
-.01485^8+3-774;

and, finally, to solve the equation for X^ the average score in
comprehension that a group of adults of limited ability prob-
ably would make if tested on the content of the book. Each
step in the procedure requires painstaking care in order that the
score thus obtained will be as accurate and reliable as possible.
A summary of data obtained by analyzing various samples of
a book for the occurrence of specific elements is necessary before
the average occurrence of each element can be determined for
the book as a whole. It seems desirable to keep a record of this

198

WHAT MAKES A BOOK READABLE

summary, together with other data used in calculating an index
of difficulty for a particular book. Such a record may be used

TABLE LV

FORM USED IN RECORDING THE OCCURRENCE OF ELEMENTS or DIFFICULTY AND
THE CALCULATION OF THE AVERAGE READING SCORE

MAKING THE MOST OF YOUR LOOKS

Author 5^*!?L?.5*i ..................... Publisher

Total Number of Pages. ...3.y. ................ Net Number of Pages. ...373.?.?.

Figured by, ......... W._D. ............................. Checked by ......... ?:A.!r.:

Chapter

VIII

XII

XIV

XVI

XIX

Page

121

153

1 86-
187

211

242

271

305

Occurrence of Elements

Aver-
age for
Book

Number of different
hard words (JQ .

21.09

Number of first-,
second-, third-
person pronouns
(X s )

10.00

Average sentence-
length in words
(Xd

I 4 .2

33-3

12.5

ii. i

I6. 7

20.26

Percentage of dif-
ferent words (X>j)

6 4

6 9

66.55

Number of preposi-
sitional phrases
(X$

8.727

later as a source of reference concerning the difficulty of a given
book, and as a means of verifying an obtained score.
Table LV is a copy of the record-card used in the present

DIFFICULTY OF ADULT READING MATERIALS 199

TABLE LV Continued
Formula Zi= .01029 X 3 + .009012 X s .02094 X 6 .03313 X 7 ~ .01485 ^8+3.774

(i) Number of personal pronouns X .009012= .090120

+ 3-774

Sum 3.864120

(2) Number of different hard words X .01029= .217016

Difference = 3.647104

(3) Average sentence- length in words X .02094= .4^4244

Difference = 3.222860

(4) Percentage of different words X .03313 = 2.204802

Difference = 1.018058

(5) Number of prepositional phrases X .01485= .129596

Final difference = Average reading score .89

study. It shows the occurrence of elements of difficulty in
State's Making the Most of Your Looks, 3 and the calculation of
the predicted score, .89. Each vertical column shows the occur-
rence of the five elements in a single sampling of one hundred
words. The general location of the sample by chapter and page
is indicated at the top of the column. Each horizontal row shows
the occurrence of a single element in the complete series of sam-
plings. At the end of each row is recorded the average occur-
rence of the element for the entire book. Space is provided on
the card for fifteen samplings, the maximum number usually
needed.

The steps followed in calculating the index of difficulty are
indicated below the tabulated elements. For the sake of ease of
computation, elements with positive coefficients are considered
first. The product of +.009012^, representing the weighted
value of personal pronouns, is added to the statistical constant,
3.774. From the positive sum thus obtained are subtracted suc-
ceeding products representing, respectively, the weighted value
of different hard words, sentence-length, different words, and
prepositional phrases. The final difference designates the aver-
age reading score or predicted index of difficulty. If these cal-

* Dorothy Stote, Making the, Most of Your Looks. New York, Brentano's, 1926.
Pp- 313.

200 WHAT MAKES A BOOK READABLE

dilations are made through the use of a calculating machine,
the process can be carried forward continuously, each answer
being kept in the machine until the final difference is obtained.
It is usually desirable, however, to record the successive an-
swers, as in Table LV, for the purpose of checking on the accu-
racy of the predicted score.

RANGE OF DIFFICULTY REPRESENTED BY 350 BOOKS

The distribution of reading scores predicted for the 350
books studied is presented in Table LVL Each interval begins
with the first number and extends up to, but not including, the
second number. For example, interval .10-. 15 includes all books
with a predicted score of .10 or more up to .15. Actually, the
end-point of this interval is .145, any score exceeding .145 being
interpreted as .15 and, accordingly, included in the next inter-
val .15-. 20.

The range of difficulty extends from 2.06, the score for the
adapted Robinson Crusoe, to .26, for the original Robinson
Crusoe. The former book is, therefore, the easiest book studied
and the latter the hardest, when difficulty is measured in terms
of structural elements. Since the highest score made by "poor-
est" readers, whose performance on the adult tests was de-
scribed in. chapter iv, was 2.1, it is evident that the adapted
version of Robinson Crusoe will probably be read with a fair de-
gree of comprehension by all adults of limited reading ability.
On the other hand, it is unlikely that any reader of the same
ability will obtain a similar understanding of the original ver-
sion.

The mean score for all books is -676. This score is representa-
tive of the difficulty ofjalna, Twelve Tests of Character, Richard
Carvel, and Rockne of Notre Dame. The standard deviation, .257,
indicates that approximately the middle two-thirds of all books
studied have a degree of difficulty varying from .933 (.676+
.257) to .419 (.676 .257), In other words, about 233 books
range between Zane Grey's Riders of the Purple Sage (.933) and
Ludwig's July '14 (.420).

DIFFICULTY OF ADULT READING MATERIALS 201

Further examination of the distributed scores shows that the
number of books in each class-interval increases with conspicu-
ous regularity from interval 1.30-1.35 to interval .65-. 70, and
decreases from the latter point with approximately the same

TABLE LVI

DISTRIBUTION OF AVERAGE READING SCORES PREDICTED FOR 350 BOOKS

Predicted Scores

Number

Predicted Scores

Number

2 OC--2 IO

80- 8<

2 OO 2 OC

. uv/~ .u^

7C- 80

I Q C 2 OO

.70- .7<

I . QO I . Q C

,6c- .70

1 . 8 C-I . QO

.60- .6c

80-1 85

* j
.r< .60

.7C i .80

. CO .

7O I .7C

.4.< . <Q

I 4

. 6 C I .70

.4.O .4<

6o~i 65 ... . .

. ^ C .4.O

I . C C I 60

.^O- .'K

I . <O I C C

,2<- .^O

I ..d.C -I CO . . . ....

.20 .2<

X "TJ * J W

I AQ I 4.C

.!< .20

I . 1 C I 4.O

.IO- .1C

I *?O I *3 C . . ...

.oc .10

T 2C I 7O

.00 .oc

I 2O I 2C

O< OO

I IC~I 2O ....

TO ,OC

I IO I I C

1C IO. .

I OC I IO

20 .1C

I OO I OC . . . .

.2C .20

orI oo

~.70 .2C

,y*j A .ww

T *7

. 90- .95

8c~ oo

*/
28

Total

TCO

Mean

6y6 .093
.066

consistency to interval .oo-.o5. The number of books at the
extremes of the scale is limited to a scattered few. What these
facts seem to indicate is that many more books of average diffi-
culty are to be found in a sampling of adult materials than are
easy or difficult books when the sampling is made as in this
study.

202 WHAT MAKES A BOOK READABLE

The distribution of books according to difficulty may be ob-
served more directly by means of the graphic presentation of
scores shown in Figure 23. Numbers below the base-line show
the class-intervals into which the predicted scores have been
distributed. High scores, representative of easy books, appear
at what is commonly termed the "low" end of scale, and low
scores, representative of difficult books, at the "high" end of
the scale. Hence, difficulty progresses from left to right, from
high to low scores.

It is at once apparent that the diagram has the same general
form as the normal curve. That is to say, we find relatively few
measures at the low end of the scale, an increasing number up to
a maximum at the mid-position, and a progressive falling-off as
we go toward the high end of the scale. If we divide the area by
a line drawn perpendicularly through the highest point to the
base-line, the two parts will be markedly similar in form. Were
the distribution perfectly symmetrical, it would show perfect
bilateral symmetry, that is, its halves would be similar in form
and equal in area. As will be shown later, it is significant for our
purposes that the distribution bears such a close resemblance to
normality.

Examination of the diagram shows that some class-intervals
have been omitted at the two extremes. Beyond interval 1.30-
1.35 at one extreme and ,00 .05 at the other, the base-line has
been shortened to include only those intervals in which fre-
quencies occur. Consequently, three easy books and one diffi-
cult book are drawn into the distribution more closely than their
scores actually warrant.

Each block on the diagram represents a book designated by a
number which corresponds to the numerical order of the book
in Table LXXXV, Appendix F. Hence, it is possible to compare
the difficulty of any one book with any other or with all others,
by identifying it on Figure 23. For the sake of illustration, let
us assume that we wish to compare Mark Twain's Tom Sawyer
with Booth Tarkington's Penrod and that we wish further to
compare both books with others along the scale. Their indexes

204 WHAT MAKES A BOOK READABLE

of difficulty, as shown in Table LXXXV, are, respectively, .86
and .71. Their positions on the diagram are indicated by num-
bers 321 and 311. It is evident that the difference in difficulty
between the two books is approximately equivalent to two class-
intervals, the former occupying the lower position on the scale.
It is equally evident that both books are easier than the ma-
jority of books studied, when difficulty is measured in terms of
structural elements.

Again, suppose we wish to determine how much variation in
difficulty has been affected by Robert Graves in his condensa-
tion of the original version of David Copperfield. In terms of a
numerical index, the difficulty has been reduced from .16 to .85,
shown on the diagram as equivalent to the difference between
numbers 92 and 91. It is clear that Graves has made from a very
difficult novel one notably easier than most of the adult books
studied.

As a final illustration, let us compare the difficulty of three
books by a single author Cimarron, Fanny Herself, and Show
Boat, by Edna Ferber. They are designated, respectively, in
Figure 23 by numbers 114, 115, and 116. As can readily be ob-
served, the difficulty of the three books is practically identical,
when measured by structural elements. While it is true that
Fanny Herself has a slightly lower position on the scale, its ab-
solute index of difficulty, .82, is not significantly different from
.78, the index for Cimarron> or .79, for Show Boat. Although
there is some reason for concluding from these facts that the
author has adopted a style of writing which represents a con-
sistent level of average difficulty, further evidence is needed be-
fore such a generalization is altogether warranted.

DEFINING AREAS OF DIFFICULTY

Having predicted the difficulty of 350 books and distributed
the scores, as in Table LVI and Figure 23, we are next concerned
with the problem of determining objectively certain areas of
difficulty by means of which the difficulty of a particular book
can be interpreted in everyday language. When its absolute

DIFFICULTY OF ADULT READING MATERIALS 205

difficulty has been predicted, we can then tell in what area of
difficulty the book rightfully belongs in relation to adult books
in general.

In the previous chapter we adopted arbitrarily three areas
"easy/' "average," and "difficult." But such areas tend to be
very broad. They do not allow fine discriminations of difficulty.
Of two easy books, one may be notably easier than the other.
Similarly, two difficult books may represent quite different de-
grees of hardness. It seems desirable, therefore, to define five
areas of difficulty designated "very easy/' "easy/' "average,"
"difficult," and "very difficult." Our problem, then, is to classi-
fy the 350 books studied into these five areas of difficulty in such
a way that the range of difficulty is equal in each.

Inasmuch as our solution of the problem involves the use of
the normal curve, probably the simplest approach is through
further consideration of Figure 23. The essential characteristics
of this diagram and its similarity to the normal curve have al-
ready been noted. Although the distribution departs somewhat
from normal symmetry, the amount of its irregularity, or "skew-
ness," is small. When the degree of skewness is stated numeri-
cally, it has a value of .023, which is interpreted as a slightly
negative skewness. 3 In other words, the scores show a slight
tendency to mass at the high end of the scale, indicated on the
diagram by low scores. When described in terms of various
measures of central tendency, the amount of irregularity is
again insignificant. The values of the mean, median, and mode,
which are equal in a normal distribution, are, respectively, .676,
.678, and .675.

If we accept this distribution of scores as sufficiently normal
to justify statistical treatment upon the assumption of nor-
mality, how can we make use of it in defining areas of difficulty?
Before answering this question we must further assume that

3 The approximate measure of skewness is found by the formula:
Hokinger, op. cft. f p. 113.

206 WHAT MAKES A BOOK READABLE

the distribution is sufficiently true to the facts concerning the
difficulty of adult books in general to make its use valid in de-
fining areas that will be applicable generally. In other words,
we must assume that a prediction of difficulty for all adult books
now available would give a series of scores whose distribution
would tend to bunch in the middle and taper off gradually
toward the extreme in a manner approximately parallel with a
normal curve. If we multiplied our 35o-book sampling many
times, and if we extended the content of the sampling to in-
clude books for all classes of readers, we would presumably dis-
cover that the general shape of the distributed scores would re-
main the same. The present difficult and very difficult books
would be drawn closer into the distribution along with many
more of their class, and a new type of difficult book would repre-
sent the upper extreme of the distribution,

While theoretically the normal curve meets the base-line at
infinite distances to the right and left of the mean, for practical
purposes the curve may be taken to end at points three standard
deviations above and below the mean. In other words, the curve
covers a range of 6<r extending from +30* to 30-. Statistical
evidence has shown that in a total of 10,000 cases, 99.73 per cent
fall within the limits set by +30- and ~yS Hence, by cutting
off the curve at these two points, we disregard only .27 of I per
cent of the distribution, an amount obviously negligible in very
large samples.

If the base-line of the distribution is taken to extend from
+30- to 3<r that is, over a range of 6<7 our problem becomes
simply a matter of defining five areas in terms of 6<r. By simple
division, we get 1.20- as the extent along the base-line allotted
to each area. These five intervals may be laid off along the base-
line, as in Figure 24, and perpendiculars drawn to demarcate the
five areas, A y B y C, Z), and E. It is clear that very easy books
(A) cover the first i.2<r; easy books (5) the next i.2<r; that aver-
age books (C) include .6cr to the left and .6<r to the right of the
mean, which is designated on the diagram 0; and that difficult

^Holzinger, op. cit. 9 pp. 113-14; aio-n.

DIFFICULTY OF ADULT READING MATERIALS 207

books (>) and very difficult books (E) occupy the same relative
positions to the right of C as B and A occupy to the left.

What do these areas mean in terms of average reading scores ?
If we obtain the average score, or index of difficulty, for a par-
ticular book, how can we tell in which area it belongs? It will
be remembered that the standard deviation (<r) has been indi-
cated in Table LVI as .257. An area of i.a<r, then, covers a
range 1.2 times .257 or .31. Knowing the mean score of the dis-

Fio. 24* Definition of five areas of difficulty represented by the adult books studied

tribution, .676, we can define the range of scores in each area on
both sides of the mean. Area C, extending .6<r above and below
the mean, includes a range of scores indicated by ,676 + .60-, or
from .53 to .84. Area B includes scores from .84 to 1.15; and
Area A, from 1.15 to 1.46 and beyond. In the opposite direction
from the mean, Area D includes scores from .22 to .53; and Area
E, from .09, or below, to .22.

With these areas defined, we can now say that Roosevelt's
Looking Forward is a book of average difficulty, for the reason
that its predicted index of difficulty, ,70, falls in Area C* School
and Home, by Angelo Patri, score i.i I, is an easy book; and the
adapted Silas Mamer, score 1.24, very easy. On the other hand,
Robinson's The Mind in the Making score .43, belongs to the

208 WHAT MAKES A BOOK READABLE

area of difficult books; and Adam's The Epic of America, score
.15, to the area of very difficult books.

It must be remembered that the assignment of these books to
their respective categories is based solely on a consideration of
one measure of difficulty, the occurrence of structural elements
bearing significant relationship to difficulty for adults of limited
ability. Unquestionably, other qualities such as informality of
style, directness of presentation, and interestingness of content
may impose a limitation on how far structural elements alone
can truly determine the difficulty of a particular book. To what
extent a more valid means of classifying books for difficulty can
be effected by a consideration of additional qualities is a prob-
lem which requires further experimentation. In the sections
which follow, it is most essential that these limitations be kept
in mind.

IS THE DIFFERENCE IN DIFFICULTY REPRESENTED BY
ADJACENT AREAS SIGNIFICANT?

If we look at this question from a statistical point of view,
the answer is "yes." The significance of the difference between
the average scores of any two books was determined from the
probable error of the difference of the scores, which was found
to be .079. Since a significant difference is usually considered
at least three times its probable error, a difference of .24 be-
tween scores on any two books may be considered statistically
significant. It is evident, therefore, that the end-score of Area
C, .53, for example, is significantly different from the opposite
end-score, .84, and hence from Area B. On the other hand, the
difficulty of the section of Area C bordering .84 is imperceptibly
different from the section of Area B bordering the same score.
In more concrete terms, The Silver Horde,, by Rex Beach, score
.82, is significantly easier than A Daughter of the Seine by Eaton,
score .55, at the opposite extreme of Area C. The difficulty of
the former is approximately the same, however, as Carroll's
As the Earth Turns, whose score of .86 ranks it in Area B; where-
as the difficulty of the latter approximates Sandburg's Mary

DIFFICULTY OF ADULT READING MATERIALS 209

Lincoln, whose score of .52 ranks it in Area D. In view of these
facts, we may say that an area of 1.20-, covering a score-range of
.31, represents a degree of difficulty significantly different in
general from any other area. Neither of its extremes, however,
is significantly different from the extreme of the area adjacent
to it.

From a practical point of view, the question of significant dif-
ferences in difficulty can be answered with considerably less
certainty. For the beginning adult reader, it seems safe to say
that a difference in scores of less than the statistical minimum
.24 may be significant. He may be able to read Country Life
Reader , Book II, score 1.33, with fair understanding, yet ex-
perience genuine difficulty with Fundamentals of Dress Con-
struction, score 1.19. His limited reading experience puts him
in the class of "poor" readers, for whom structural elements, as
shown in chapter iv, are notably more closely related to diffi-
culty than for "good" readers. For the latter type, that is, for
the person with mature reading habits, the ranking of a book as
"easy," "average," or "difficult" is relatively unimportant.
From the point of view of structure, probably no general book
is so difficult that he cannot read it, should he choose to do so.
Whether he will choose to read a very difficult, an average, or
an easy book is another matter.

WHAT BOOKS CHARACTERIZE AREAS A, B, C, D, AND E?

In characterizing areas of structural difficulty in terms of
specific books, we have attempted to select from among the 350
books studied those which are familiar to a large number of
readers of this report. With each book cited in this section are
given its index of difficulty, or predicted score, listed in Table
LXXXV, Appendix F, and its numerical order in that table by
which it may be identified in Figure 23.

Of the 350 books studied, 10 belong to Area A. In other
words, their index of difficulty is higher than 1.15, giving them
lowest rank along the scale of difficulty. Two types of books
were found to fall in this area of very easy reading* One type

WHAT MAKES A BOOK READABLE

includes simplified materials prepared especially for beginning
adult readers and near-illiterates. The other includes simple
books on practical subjects written to reach a large reading
audience. Among the former are the Country Life Readers,,
Books I and II, numbered, respectively, 301 and 302; Home and
Health in a New Land (No. 129); and adaptations of Silas Mar-
ner (No. 194) and Robinson Crusoe (No. 334). The scores of
these books range between 1.24 and 2.06, indicating that adults
of limited ability can probably read them with a fair degree of
understanding. Very easy books on practical subjects include
Brigham's Box Furniture (No. 47), score 1.24; Manning and
Donaldson's Fundamentals -of Dress Construction (No. 214),
score 1.19; and Maternity Handbook (No. 223), score 1.16.

Area B extends from Farm Blacksmithing, by Friese (No.
126), to the condensed edition of David Copperfield (No. 91).
The predicted score of the former is 1.13; of the latter, .85.
Despite the fact that both belong to the area of easy books, the
one is about as closely related to the hardest of very easy books
as the other is to the easiest of average books. Among the 70
easy books are found School and Home y by Angelo Patri (No.
252)3 score i.ii; Ilin's What Time Is It? (No. 171), score 1,04;
and New Russia's Primer (No. 170), score .97. More books on
practical subjects making of draperies, growing house plants,
caring for the home, designing furniture, repairing shoes are
among the books which adults of limited ability presumably find
easy. A few popular novels rank in this area. Conspicuous
among them are Zane Grey's Riders of the Purple Sage (No.
139) ; A Lantern in Her Hand, by Bess Streeter Aldrich (No. 9) ;
and The Bonney Family, by Ruth Suckow (No. 858). The score
of the first is .92; of the other two, .86. This latter score is repre-
sentative also of Mark Twain's The Adventures of Tom Sawyer
(No. 321), and Mary Antin's The Promised Land (No. 16).

In Area C, extending from score .53 to score .84, popular
novels and popularly written general non-fiction predominate.
Here we find a variety of good stories, both old and new:
Dumas, The Three Musketeers (No. 98); Tracy, Wings of the

DIFFICULTY OF ADULT READING MATERIALS 211

Morning (No. 316); De La Roche, Jalna (No. 269); London,
The Call of the Wild (No. 203); Ferber, Cimarron (No. 114),
Fanny Herself (No. 115), and Show Boat (No. 116); Lewis,
Babbitt (No. 196); Bromfield, The Green Bay Tree (No. 48);
Deeping, Sorrell and Son (No. 87); Buck, The Good Earth (No.
51); O. Henry, Four Million (No. 151); Hey ward, Mambas
Daughters (No. 157); Wharton, Ethan Frome (No. 336);
Hough, The Covered Wagon (No. 165), and many other stories.
They represent adventure, mystery, family life, and characteriza-
tion. They are stories of the North, East, South, West, and
Middle West. In one respect these stories are all similar. They
are told in a manner which from the point of view of structural
elements marks them of average difficulty among adult books
in general.

In this "average" area, too, we find a generous amount of
biography: Lindbergh, We (No. 199); White, Daniel Boone
(No. 337); Finger, David Livingstone (No. 112); Partridge,
Amundsen, the Splendid Norseman (No. 2,51); Repplier, PSre
Marquette (No. 262); Sandburg, Abe Lincoln Grows Up (No.
281); Lovelace, Rockne of Notre Dame (No. 205); Cantor, My
Life Is in Your Hands (No. 59); Barrymore, Confessions of an
Actor (No. 27); Winkler, John D: A Portrait in Oil (No. 342);
and Earhart, The Fun of It (No. 99), These few titles by no
means exhaust the list of interesting life-stories of interesting
people, told by themselves or by others in such a way that read-
ers of limited ability may be expected to find them of average
difficulty.

Books on sports generally rank in Area C. That is, their pre-
dicted index of difficulty falls between .53 and .84. Among such
books we find: Barnes, Swimming and Diving (No. 24) ; Moore,
The Mental Side of Golf (No. 237) ; Hulit, The Salt-Water Angler
(No. 169); Hammett, Major Sport Fundamentals (No. 143); and
Lacoste on Tennis (No. 189).

A more restricted number of books of history rank in Area C.
History in general is represented by Parson, The Stream of His-
tory (No. 250); Hartman, These United States (No. 146); Herd-

2i2 WHAT MAKES A BOOK READABLE

man, History of the United States (No. 153); and Singmaster,
The Book oj the United States (290). Accounts of the World War
include Remarque, All Quiet on the Western Front (No. 261) ; and
Masefield, Gallipoli (No. 220). Books of history, in general, do
not fall among "average" books. They tend, rather, to belong
to Area D, which represents what we have termed "difficult"
reading materials.

In Area D, extending from score .22 to score .53, are included
69 books which from the point of view of structure are probably
too difficult for adults of limited ability to comprehend satis-
factorily. This does not mean that so-called "difficult" books
cannot be read understandingly by some relatively inexperi-
enced readers. Undoubtedly they can and for the reason that
other factors may offset structural difficulty. For example, a
book of compelling interest with an index of difficulty of .35 may
be more intelligible than another with an index of .68. In gen-
eral, however, Area D is characterized by books of a level of
difficulty too advanced for readers of limited ability.

Few novels among the 350 books studied rank in Area D.
These few are representative of the more substantial type of
fiction novels of the Victorian Era, ponderous novels equal to
two or three of ordinary length, and European novels in trans-
lation. Among them we find: Austen, Pride and Prejudice (No.
19); Goldsmith, The Vicar of Wakefield (No. 135); Turgenev,
Fathers and Sons (No. 320) ; Rolland, Jean-Christophe (No. 272) ;
Balzac, Pere Goriot (No. 22) ; and Daudet, Tartarin of Tarascon
and Tartarin on the Alps (No. 86).

Like fiction, biography is represented relatively little in Area
D. Illustrative of difficult biography are: McMahon, Wright
Brothers (No. 228); Anderson, These Quarrelsome Bonaparte s
(No. u); and Arliss, Up the Years from Bloomsbury (No. 17).

What sort of books, then, best characterize Area D? Two
kinds rank most frequently as difficult reading. They are books
of travel and books of history. The former include: Mason,
Columbus Came Late (No. 222) ; Leys, After You, Magellan! (No.
197); Van Loon's Geography (No. 326); Duguid, Green Hell (No.

DIFFICULTY OF ADULT READING MATERIALS 213

97) ; Woolley, Ur of the Chaldees (No. 346) ; Morton, In Search of
Scotland (No. 239); and Banks, The Story of Mexico (No. 23).
Difficult books of history are represented by Davis, Life in
Elizabethan Days (No. 85); Ludwig, July '14 (No. 209); Gibbs,
Since Then (No. 132); Lang, The Conquest of Montezumas Em-
pire (No. 191); and Maurois, Disraeli (No. 224). That travel
and history should present difficult reading for the limited read-
er is not wholly unexpected. Much of the difficulty can un-
doubtedly be ascribed to the frequent occurrence of proper
words which tend to weight the vocabulary with hard words.

Thirteen of the 350 books studied have an index of difficulty
below .22. They are the very difficult books which represent
Area E. Among them we find: Allen, Only Yesterday (No, 10);
Mowrer, Germany Puts the Clock Back (No. 240) ; Adams, The
Epic of America (No. 3) ; Rogers and Allen, The American Pro-
cession (No. 271); Sullivan, Our Times, 1900-1914 (No. 308);
and Bowers, The Tragic Era (No. 42). Curiously enough, these
"very difficult" books deal with subjects of wide appeal. They
present some vital problem of present-day living or some aspect
of an earlier period that has made appreciable contribution to
our times. They are written in a vivid, narrative style that is
easy and agreeable for the able reader. Yet from the point of
view of structure they present difficulties which put them be-
yond the understanding of readers of lesser ability. How much
this handicap can be overcome by the reader's attitude, the
purposiveness of his reading, the attractiveness of the theme,
and other potent factors of the reading situation, is a question
that cannot be answered. Undoubtedly, such factors tend to
decrease appreciably the influence of structural elements on
difficulty.

WHAT DO AREAS OF DIFFICULTY MEAN IN TERMS OF
GRADE LEVELS OF READING ABILITY?

More specifically, does Area C represent a degree of difficulty
appropriate for grades 4 and 5 as for grades 7 and 8 ? Can adults

WHAT MAKES A BOOK READABLE

whose reading ability is defined at the sixth-grade level read
understandingly difficult books ranked at Area D ?

In order to answer these questions precisely we need to find
out how well adults of different degrees of reading ability can
comprehend materials at Areas A, B, C, D, or E. If by testing
their understanding of a variety of materials, we should find
that the majority of adults of a given grade comprehend "aver-
age" (Area C), but not "difficult" books (Area D), then we
might say that materials at Area C are best suited to their level
of ability.

A comprehensive testing program, however, meets a number
of practical obstacles. In the first place, although a large per-
centage of adults engage in some reading of a general sort, a
relatively small percentage attend organized classes of instruc-
tion. Then, too, the few adults who attend day or evening
schools are seldom classified formally at a specific grade-level.
To interpret areas of difficulty in terms of adult grades of edu-
cational progress is, therefore, not only an intricate task but one
that is more or less impractical. On the other hand, if we inter-
pret their meaning in terms of elementary grade-levels for which
they are structurally appropriate, we shall have an estimate of
difficulty that is generally intelligible and presumably practical.

Considerable effort has been centered on determining the ap-
propriate grade-placement of reading material for children.
This is largely for the reason that fundamental reading habits
show progressive growth throughout the elementary grades
until they reach a stage of maturity normally attained by the
end of the elementary school. In harmony with these facts,
ability in reading has been accepted as a fairly valid index of a
child's ability to meet the requirements of a particular grade.

The careful grading of children's reading materials has con-
sequently been essential in order to provide for pupil adjustment
at each grade-level and for pupil advancement from one grade-
level to the next. Materials have been graded on the basis of
such factors as familiarity of vocabulary, length and complexity
of thought-statements, number and quality of ideas, and ap-

DIFFICULTY OF ADULT READING MATERIALS 215

propriateness of content to the major interests at any given
level. As a result, we now find many textbooks for each grade
prepared with such care that they suit the reading needs and
abilities of the majority of pupils classified in that grade.

In attempting to establish a meaning of difficulty for adult
materials in Areas A, B, C, D, and E we have compared the
indexes of difficulty represented by each area with the indexes of
difficulty of a wide sampling of children's textbooks in reading.
It is not our purpose to conclude from such comparisons that
the difficulty of adult books in any given area is equivalent to
that of children's books at a given grade-level, save with respect
to structural elements.

Eighty-one textbooks in reading were chosen for study. In-
cluded in this number are eleven series of readers for grades 2 to
6, inclusive, and twenty-six widely used texts in reading and
literature for grades 7 and 8 of the elementary school or for
grades 7, 8, and 9 of the junior high school. The books are
listed in Table LXXXVI, Appendix G, together with the name
of author and publisher, and date of publication of each.

The distribution of average reading scores predicted for the
eighty-one textbooks for grades 2 to 9 is shown in Table LVIL
The range of difficulty extends from 1.60, the score predicted for
The Open Door, a reading text for second grade, to .16, the score
for Literature and Life, Book II, a text for eighth grade. The
table shows interesting differences between the difficulty of
books at successive grade-levels up to the period of the junior
high school. These differences are more strikingly presented in
Figure 25. On this diagram are shown the range of difficulty
among the books for each grade, the average difficulty, and the
range of difficulty represented by the middle two-thirds of books
studied for each grade.

Up to the eighth grade, books for each grade show an average
difficulty that is notably higher than for the one immediately
preceding. The average difficulty of books for eighth grade is
but slightly higher than for seventh grade, and but slightly
lower than for ninth grade. Although there is a degree of over-

2l6

WHAT MAKES A BOOK READABLE

lapping in the range of difficulty of books for succeeding grades,
it is not until the junior high school that the range of difficulty
for one grade closely parallels the range for the next.

TABLE LVII

DISTRIBUTION OF AVERAGE READING SCORES PREDICTED FOR
EIGHTY-ONE TEXTBOOKS IN READING

Predicted
Scores

All
Book

Sec-
ond
Grad

Third
Grade

Fourth
Grad

Fifth
Grad

Sixth
Grade

Sev-
enth
Grade

Eighth
Grade

Ninth
Grade

i 60 i 6c

T t f T 60

1 -5-> A >ou

I CO I CC

* jr^*- 'jj

1 .4^) A .^)U
I A.Q- 1 AC

T 7 C I A.Q

I QO-I TC

1 jr* -Ji

I 1 C I ^O

1 "*J a *J W
I 2O~I 2 C

I 151 20

-3

I IO I 1C

i 051 10 . . ...

I.OO-I .O^

-3

Q C I OO

QO~ QC

yv^ -yj

.8c- QO

80- . 8 <

7C- 80

7O- .7C

/^ /:>
.6<- .70

,J /
. 60- . 6 <

v ;.J

.c<- .60

. 0- , C C

.4.C- . CO

,4.0 .4.C

,TC- .40

.30- ,35

.2C- .70

.20- .2-5

.1C- .20

Mean

T .^

i 18

I OO

QOQ

772,

616

6oA

/7*7

Probable error. ...

.016

4- OOQ

01*1

.yvjy
.OIC

' // t

i ooo

OI C

<W4

db 021

577

OI C

0_ ,wwy

.WAJ

.wij

Standard deviation
Probable error. . . .

.147
db .011

.087
006

.122

4- OQ2

133

i on

.085

OOO

135

i oil

.192

01 1

.133

i 01 1

j_ .wy*

DIFFICULTY OF ADULT READING MATERIALS 217

The inference to be drawn from these facts is that whereas
structural elements tend to indicate important differences in
difficulty of reading materials throughout the elementary school^
some other factors are essential in measuring difficulty at the
junior high-school level and presumably beyond. Such an
inference does not mean to imply that other factors are of no

Score

.00

. 4

r-i

50
.6o

- '

.70
,80

.90

J ^- -

1.00

1.20

t-30

- -i<r

1.40

I.SO

1.60

_ -Her

II III

IV V VI

VII VHI IX Grade

FIG. 25. Predicted difficulty of reading texts for elementary and junior high schools
in terms of range, mean, and standard deviation.

consequence in determining difficulty in the elementary grades,
but that sufficient difference exists with respect to the occur-
rence of structural elements to make their use possible in esti-
mating difficulty. Whether these other factors are related to
number and quality of ideas, to the degree of directness by
which the ideas are presented or to some other aspect of style,
or whether they are related to the breadth and depth of the
reader's experiences are questions which require further study.

ai8

WHAT MAKES A BOOK READABLE

How can the facts presented in Table LVII and Figure 25 be
used in interpreting the difficulty of adult reading materials
characterizing Areas A, B, C, D, and E? Figure 26 suggests the
answer to this question. In the diagram is shown the distribu-
tion of scores of adult books as in Figure 23. Areas of difficulty
are indicated along the base-line by the sigma values defining
each. Superimposed on the distribution are rectangular blocks

Middle two-thirds of

grade textbooks
Below middle two-thirds
Above middle ewo.th.irds

Grade

-.00

Area A

AreaB

ArcaC

Area D

Area E

Fro. 26. Interpretation of areas of difficulty in terms of difficulty represented by
textbooks for grades H to IX.

illustrating the same facts that appear in Figure 25. The length
of each block represents the range of scores predicted for all
books for a given grade. The shaded portion indicates the range
of difficulty of the middle two-thirds of the books studied. The
curve through the blocks shows the trend of average difficulty of
books for grades 2 to 9, inclusive.

When areas representing the difficulty of adult books are
compared with areas of difficulty for children's textbooks in
reading, the following conclusions seem warranted:

i. Least difficult adult books, that is, "very easy" books in
Area A, are about as difficult from the point of view of struc-
tural elements as reading textbooks for second grade and most
reading textbooks for third grade. If the 350 books studied may
be taken as representative of adult books in general, it is evident

DIFFICULTY OF ADULT READING MATERIALS 219

that relatively few fall at this level. These few are simplifica-
tions of standard novels and other purposeful attempts to ex-
press adult subject matter in very simple language.

2. The difficulty of "easy" adult books in Area B is best de-
scribed in terms of reading textbooks for fourth grade. Approxi-
mately all the texts for this grade have indexes of difficulty be-
tween .84 and 1.15. A large number of texts for fifth grade and
a considerably smaller number of third-grade texts also rank in
Area B.

3. Area C, containing "average" adult books, represents a de-
gree of difficulty equivalent to that of most reading texts for
sixth grade and of many readers and literary texts for junior
high school. Although the mean difficulty of books for eighth
and ninth grades falls within this area, a considerable portion of
the total range of difficulty represented by these books falls be-
yond Area C into Area D. Inasmuch as a large number of books
on a variety of subjects may be characterized "average," it is
evident that much adult reading material is no more difficult,
structurally, than the average reading text for sixth grade and
junior high school.

4. "Difficult" books in Area D are less easily described in
terms of grade levels. If we say that they are equivalent, from
the point of view of structural difficulty, to many reading texts
for junior high school, we fail to tell the whole story. They are
that and more. If we assume that they are equivalent in diffi-
culty to texts for senior high school, we shall find that evidence
in support of the assumption is not easily obtained. Textbooks
in reading are not usually provided for this level, and other
books tend to deviate from the general into some speciali2ed
field. Were such facts obtainable, we should still be forced to
question whether they present a true picture of difficulty at this
level, where structural elements are of less significance,

5. The meaning of "very difficult" books cannot be interpret-
ed with precision from data in Figure 26. That Area E is out-
side the range of difficulty represented by the textbooks studied
is obvious. Structurally, then, it represents a degree of difficulty

220 WHAT MAKES A BOOK READABLE

beyond that of reading texts for elementary and junior high
schools. Perhaps, however, at this area, structural difficulty is
relatively less important than other qualities in determining
whether a reader who is able to read a book classified in Area D
will meet any more serious difficulty in Area E. It seems alto-
gether probable that the structural obstacles in one will be no
more significant than in the other, inasmuch as data already
presented show that structural elements are less closely related
to difficulty for "good" readers than for "poor" readers. On the
other hand, at this area, as elsewhere, structural elements are
of considerable significance for readers of limited ability, al-
though vital interest in the content of a very difficult book and a
genuine motive for reading may reduce its difficulty as defined
by structural elements alone.

WHAT PERCENTAGE OF ADULT READERS CAN READ BOOKS
AT AREA A, B, C, D, OR E?

Since the general notion of differences in difficulty of reading
materials implies differences in reading ability, it appears that
the findings of the present chapter may gain in meaning if
viewed in the light of the findings of chapter iii. It will be re-
called that about one-sixth of the adults tested were found to
read with a proficiency normally attained by high-school gradu-
ates. These are the readers, then, who can presumably read
any general book with an index of difficulty falling at any point
along the scale from Area A to Area E. This need not imply
that such readers have acquired mature and unmodifiable read-
ing habits nor that their difficulty in reading is unrelated to
structural elements. On the contrary, investigators have re-
cently found gains in various aspects of reading ability attribut-
able to training even at the college level. Furthermore, data in
chapter iv give proof that structural elements may influence
difficulty in some degree for the best readers. What seems ten-
able to conclude, therefore, is that although one-sixth of the
population studied is presumably able to read general material
even of a very difficult sort, they may find technical material,
especially in an unfamiliar field, difficult to comprehend.

DIFFICULTY OF ADULT READING MATERIALS 221

Evidence presented in chapter iii shows further that approxi-
mately half the adults tested made a reading score equivalent
to grade 7.0 or above. If the group tested may be taken as a
fairly representative sample of the total population, then we
may conclude that about half are able to read understandingly
most of the material in Area C. And they have a vast amount
of interesting material to read. It includes a variety of good
stories and informational books on travel, biography, sports,
and so on, the difficulty of which falls well within Area C, the
realm of "average" books.

In view of the fact that about one-third of the adults co-oper-
ating in the study have a reading achievement at fifth grade or
below, a large number of readers cannot be expected to read
understandingly materials of greater difficulty than those in
Area B. This conclusion should probably be qualified some-
what, for the reason that readers of lesser ability may read a
novel or biography from Area C with emotional satisfaction, if
not with complete comprehension. In such case, we may ques-
tion whether he will read books below this area with the same
degree of pleasure. In the present study we have made no at-
tempt to measure emotional outcomes of reading, save as they
be measured indirectly with comprehension.

The findings of chapter iii indicate that about one-sixth of the
adults tested have a reading ability equivalent to the norm for
fourth grade or below. For them, materials in Area A and Area
B seem appropriate. But such materials are relatively few in
number. Accordingly, a large percentage of adults whose educa-
tional progress depends on extensive reading of simple, easy
material must at present gather their reading experiences large-
ly from children's books.

What we have just attempted to show is that large differences
in the difficulty of adult books are paralleled in a degree by dif-
ferences in adult reading ability. For every general reader, then,
there is some reading material now available which he can read,
provided that we can get book and reader together. How this
objective may be accomplished by a knowledge of the difficulty
of a particular book is what we propose to show in the chapter

222 WHAT MAKES A BOOK READABLE

which follows. The amount of reading material of appropriate
difficulty is not equal for various levels of ability. For one class
of readers there is an abundance of suitable material; whereas
for another class simple, easy materials are meager in quantity.
To provide adult books simple enough to fit readers of low abili-
ty is a problem for the writer and the publisher. This problem
we shall consider in a later chapter.

CONCLUDING STATEMENT

The facts presented in this chapter indicate that differences
in the occurrence of significant structural elements may be
utilized in comparing the difficulty of adult reading materials.
The use of these elements in obtaining an index of difficulty for
specific books has been illustrated to the point of showing, first,
that the structural difficulty of adult books in general covers
a wide range; and, second, that significantly different areas of
difficulty may be defined between the most difficult and the
least difficult book in the total range.

By a definition of these areas, it becomes possible to speak
relatively yet precisely concerning the difficulty of a given book
relatively, because an "easy*' book is easy in relation to other
books; and precisely, because the relative designation "easy"
is merely a descriptive interpretation of an accurate, numerical
index. By interpreting each area concretely, we are able further
to describe a given book as one that is about as difficult struc-
turally as textbooks in reading for a particular school grade.

Although we have classified books along a scale of difficulty
from "very easy" to "very difficult," it must be remembered
that this classification is intended only for readers of limited
ability, that is, for persons who, like our "poorest" readers, find
in general materials structural obstacles to understanding. The
classification would be quite different were it made applicable
to adults whose reading ability equals or exceeds our "best"
readers. For such individuals, The Epic of America, Only Yes-
terday, The Tragic Era, and other of the "very difficult" books
for limited readers would probably be classified "easy," if not

DIFFICULTY OF ADULT READING MATERIALS 223

"very easy." At higher areas would be ranked another sort of
"difficult" book, such as Eddington, Nature of the Physical
World; Haldane, Daedalus , or Science and the Future; Newman,
Nature of the World and Man; and Spengler, Decline of the West
books in which difficulty is influenced more by the quality of
the concepts presented than by inherent elements of expression.
In order to classify books for the highly skilled, independent
reader, we need a new instrument of prediction based upon new
criteria of difficulty and upon inherent elements other than
those of structure. Whether a classification of books for these
independent readers is practicable or necessary is a problem
that merits investigation. For all classes of readers we need
ultimately a more refined instrument of prediction, one that will
recognize the interrelation of all factors of the reading situation.
And this interrelationship can be determined only by extended
study of the kind mentioned throughout this report.

CHAPTER VII

HOW TO SELECT READING MATERIALS
FOR ADULTS

HOW to select readable materials for adults is the practi-
cal consideration of this chapter. We have approached
the problem in the light of the following questions:
What is its importance for advisers and teachers of adults?
What are the issues which must be considered in selecting read-
able books ? What is the value of information concerning ele-
ments of difficulty in selecting reading materials ? How can such
information be put to practical use by librarians, readers' ad-
visers, and teachers of adults ?

Although we have attempted to discuss these questions in
considerable detail, we have not attempted to answer them.
That is more than the scope and accuracy of the present study
will warrant. In succeeding sections of this chapter discussion
of the foregoing questions aims to do two things: first, to stimu-
late thoughtful consideration of the important problem of se-
lecting reading materials for other persons to read; and, second,
to illustrate how the findings of objective studies may be applied
in practical situations. What we propose in the latter connec-
tion is not an entirely new method as a substitute for current
practice but, rather, a means of unifying, objectifying, and sup-
plementing methods now in use through an application of the
findings presented in the preceding chapters.

IMPORTANCE OF THE PROBLEM

Guiding adult readers in a more effective use of printed ma-
terials is no new responsibility for the librarian. It is, however,
a greater responsibility than it was when the amount of library
reading was notably less than at the present time. Reports from
sample libraries in thirty-three cities show that the amount of

224

HOW TO SELECT READING MATERIALS 225

reading since 1929, as measured by the number of volumes bor-
rowed for home reading, has increased more than 50 per cent for
seven cities reporting. 1 For Hammond, the increase is 173 per
cent; for Akron and Dallas, 116 per cent each; for Dayton, 67
per cent; for Washington, D.C, 64 per cent; for Memphis, 59
per cent; and for Houston, 58 per cent.

An examination of library records for other cities not included
in the general survey reveals the same phenomenal growth dur-
ing the period of the depression. Circulation figures at the
Evanston Public Library, for example, have increased 67.44 per
cent since I928. 2 More than 50 per cent of the residents of
Evanston were accredited borrowers at the close of 1933, repre-
senting an increase in number of patrons amounting to 21.6 per
cent since 1929. Remarkable as these figures are, they fail to
give a true picture of the growth in library patronage during the
five-year period, for the reason that only the reading of ac-
credited borrowers is recorded. There still remain to be con-
sidered scores of persons everywhere who come to the public
library to read newspapers, periodicals, and books of reference,
but who have no cards and who never make withdrawals.

Increased library patronage finds ready explanation in the
present economic situation. Persons who normally spend their
leisure in other forms of recreation are now turning to the li-
brary in steadily increasing numbers. Some come to "browse,"
because there is nothing else for them to do. Others come with
the purposeful intent of seeking security, comfort, and happi-
ness in self-improvement. These last are the readers "with a
purpose" who find in reading a new avocation; or who aim to
gain new knowledge about trades and professions of which they
are relatively uninformed; or who hope to find some explanation
for current social and economic changes. They are the readers
of non-fiction the serious readers who are pursuing definite
courses of reading on general or specialized subjects,

1 "Increase in Reading since 1929, Shown in Reports from Sample Libraries in 33
Cities," Library Journal, LVIII (January 15, 1933), 77.

a "Public Keeps Pace with New Conditions," Evanston Review, IX (February 22,
1934).

226 WHAT MAKES A BOOK READABLE

Despite the fact that many persons come to the library daily
without seeking guidance from the library staff, others are regu-
larly requesting such service. They represent an interesting va-
riety of readers: the white-collar class students, business men
and women, and professional people the laboring group, and
the endless army of the unemployed. Probably the library itself
has encouraged requests for advisory service by its growing
tendency to provide it. Whether the one is the cause and the
other the effect, or vice versa, is relatively unimportant. The
significant fact is that adults are increasingly seeking guidance
in their reading and that libraries are increasingly providing it.
In the larger libraries where the need for guidance is particularly
acute, skilled advisers and consultants have been added to the
staff to promote more intimate contact between library and
reader, to diagnose the readers' needs, and to furnish reading
courses to fit these needs. In 1929, some twenty-five libraries
had one or more specialists to direct adult reading. This num-
ber since has increased to forty-four. Other libraries, the num-
ber of which cannot be estimated, are instituting "made-work"
programs and CWA projects designed to effect such changes in
the routine work as will enable regular librarians to devote more
time to advising adult patrons.

Advisory activities themselves are undergoing significant
changes. Generalized forms of guidance are giving place to more
particularized attempts to fit reading materials to individual
needs and capacities. Lists of books on various subjects, pam-
phlets and bulletins orienting a given field, and similar guides
for the general reader are being increasingly supplemented by a
more personal kind of assistance. It is personal because it is
based upon reliable information concerning a reader's particular
interests, his educational background, and his social and eco-
nomic status. Every effort is directed toward fitting each in-
dividual's reading to his interests and abilities so effectively
that it will yield him the greatest possible good.

The problem of selecting appropriate reading materials for
adult readers is not confined alone to voluntary reading of the

HOW TO SELECT READING MATERIALS 227

"out-of-school" sort. It seems equally acute in connection with
the more formalized type of reading provided for instructional
use in school. Unfortunately, many adults cannot pursue self-
education unaided. Among them are foreigners, illiterates, and
others who for one reason or another missed early educational
opportunities or who are being forced to make an immediate
adjustment to vocational demands.

In addition to these adults, "there are always to be found al-
most numberless persons in whom valuations, ideals, and
trained powers of self-education are needlessly low, due perhaps
in part to failures of previous school educators and perhaps
those of other agencies to hold the important goals of self-
education in view as among desirable objectives/' 3 They meet
in evening schools, community centers, church schools, classes
for the unemployed, or in other of the numerous agencies apply-
ing the traditional instructional technique for promoting edu-
cational progress. Others are to be found in prison schools and
in classes in county jails, where the work of education and cor-
rection go hand in hand.

An analysis of activities engaged in by adult-education agen-
cies indicates that those of a cultural nature are now exceed-
ing all others. Relatively few students are in school to satisfy
a casual interest. Many are demanding a continuation of gen-
eral education and are manifesting a vital interest in intellectual
advancement. Some reading materials must therefore be pro-
vided. More must be recommended. Their selection must aim
to fit the needs and abilities of students at various stages of
progress. The obligation placed upon teachers and directors is
a serious one. It is serious, first, because the evidence seems
conclusive that there is a dearth of good readable material for
adults of limited ability; and, second, because an objective tech-
nique has not been developed for determining with accuracy
what material is readable for what persons.

Selecting appropriate reading matter for adults is a problem

* David Snedden, "Self-Education: A Needed Emphasis on Current Proposals for
Adult Education," Journa I of Adult Education, II (January, 1930), 32-37.

228 WHAT MAKES A BOOK READABLE

that extends to such educational agencies as forum, radio, and
others which enable people to keep on learning while being en-
tertained. Inasmuch as the forum aims to promote a broad un-
derstanding of current problems and a critical attitude toward
them, the value of recommended readings is obvious. In some
communities,, one member of the library staff is in constant con-
tact with the forum meetings. Through co-operation with the
forum leaders, this librarian prepares book-lists and bibliog-
raphies on numerous subjects. These are discussed and distrib-
uted by the leader at the evening meeting. 4 In other commu-
nities, the attempt to prepare book-lists fitted to the needs of a
heterogeneous audience has come to be recognized as futile.
Instead, individual book-lists which aim to fit the reading needs,
interests, and abilities of a particular reader are prepared upon
request. What technique to employ in order that such informa-
tion may be secured expeditiously and book selection be made
effectively is a problem of no small importance in reading ac-
tivities related to the forum.

With respect to education on the air, it is evident that no
other educational medium can equal the radio in reaching large
audiences. Whether the members of these audiences listen and
learn or merely listen is a perplexing question. It is believed,
however, that the chances of the first alternative being achieved
are multiplied by a properly organized follow-up program. To
this end, the National Advisory Council on Radio in Education
has adopted two methods by which to guide listeners in follow-
up activities. 5 The first is to distribute printed material among
the radio audience in the form of a "Listener's Manual" or note-
book. Each notebook contains a foreword or brief introduction
to the field in which a particular series of lectures is delivered.
Summaries of the lectures are then presented, followed by ques-
tions for group discussion. Finally, there appear lists of sug-

Mildred 0. Peterson, "Des Moines Holds Public Forums/' Library Journal,
LVHI (May 15, 1933), 453-54.

* Levering Tyson, Radio and Education. Proceedings of the Second Annual Assem-
bly of National Advisory Council on Radio in Education. Chicago: University of Chi-
cago Press, 1932. Pp. 306*

HOW TO SELECT READING MATERIALS 229

gested readings for beginners and for more advanced students.
The second method in the follow-up program is provided
through co-operation with the American Library Association,
which arranges to have books on the collateral reading-lists
made available in libraries.

An evaluation of these methods shows them to be far less suc-
cessful in interesting the radio listener to read than is commonly
expected. Probably the reading-lists themselves are partly at
fault. In reporting to the Advisory Council in this connection,
the official radio publishers make the following statement: "Fu-
ture printed material for listeners should be even more carefully
prepared and should go much farther in stimulating independ-
ent reading and use of the information received during the broad-
cast* More attention should be given to the gradation of the
material in each series, and to tests and other checks to record
the listener's progress and to stimulate him through the thrill
which comes from knowledge of such achievement." 6

Obviously, there is no typical radio listener, any more than
there is a typical reader. If listeners are also to be readers of the
subject in which they are interested, recommended readings
must be appropriately graded from easy, short, general pres-
entations of the subject (if there are such) for the least compe-
tent reader to detailed, technical treatments for the scientific
reader.

The problem of selecting readable books is relatively less im-
portant when viewed in its relation to the home consumer the
reader who is also the buyer of the books he wants to read. For
the purchasing of books does not loom high in the expenditures
of American families. According to figures for commodity sales
presented by the Census of Distribution, "the proportion of
America's dollar going to the bookstore is about one-fifth of a
cent, as compared with 19 cents to the automobile industry, 1.4
cents to the candy store, one-third of a cent to the florists, and
i.i cents each to jewelry stores and radio and music shops." 7 A

6 Ibid., p. 78.

7 CX H. Cheney, Economic Survey of the Book Industry, fgjo-r^jf (New York: Na-
tional Association of Book Publishers, 1931), p. 56.

230 WHAT MAKES A BOOK READABLE

detailed investigation of the living standards and expenditures
of one hundred families of employees of the Ford Motor Com-
pany in Detroit shows that of the total budget only one-hun-
dredth of one per cent was spent on books. 8 Increased income
means little difference in the buying of books. Bigger shares of
larger incomes go to automobiles, movies, travel, vacation but
not to books.

How many book-buyers know exactly what book they want
to purchase when they enter a bookshop is not known. That
many examine a book critically before purchasing in order to see
how nearly it suits their individual needs and interests is a fact
to which most of us bear witness. Such persons can make their
own selection if the right book is available. Others are more de-
pendent upon salesmen whose information about books and read-
ers is their best means of bringing book and buyer together. Any
improvement in methods of obtaining either type of information
presumably will tend to increase book sales and subsequent
book-reading.

The selection of the right material for the right reader, then,
is a problem of no small consequence. Recreation for leisure
hours, intellectual progress, even literacy itself, depends in a
large measure on how well the problem is solved by the reader
himself or by someone from whom he seeks advice and counsel.

WHAT ASPECTS ARE TO BE CONSIDERED IN
SELECTING READABLE MATERIAL?

Selecting a readable book, like defining a readable book, is a
highly individual problem. It depends, first, upon the reader's
interests, needs, and abilities; and, second, upon the qualities of
a book that make it readable for him. Information of both sorts
must be obtained by the person who would guide a reader most
wisely. Studies of advisory practices in libraries indicate that
many different kinds of information are obtained by advisers as
a group. Some advisers maintain a traditional conservatism to-
ward Inquiring into the "ins and outs" of a person's history.

. 58.

HOW TO SELECT READING MATERIALS 231

Others proceed frankly and directly to enlist a patron's co-oper-
ation in determining all factors that may influence his reading
needs.

Winslow has discovered, from replies of eighteen readers' ad-
visers to a questionnaire, that information on seventy-two items
is being secured more or less frequently by at least one adviser. 9
The purpose of this information is indicated by the following
statements: to determine specific interests and problems, to
evaluate a patron's judgment and tastes, to determine the in-
tensity of his reading interests, to determine his ability and dis-
ability, to determine external obstacles to reading, to evaluate
books and reading courses for a patron, and to determine a form
of study best suited to his needs. The composite list of items
gathered by Winslow include: age, sex, place of residence, na-
tionality, length of residence in the United States, amount and
nature of previous schooling, formal classes now being attended,
personal traits aiding or handicapping the student, occupation,
vocational ambitions, recreations and hobbies, books read in the
past year, most interesting book ever read, subject of reading
courses requested by the patron, reasons for electing the specific
course, and so on, to a total of seventy-two.

Not all librarians record the information obtained from their
patrons. "To some librarians," Mason observes, in commenting
on library practices, "records are vital; to others, they are
negligible. Some seem to consider the records ahead of the read-
er and his needs. Others fear that the mechanics will overshad-
ow the personal equation and they neglect the great possibilities
of service that records might render." 10 If information about a
reader has no value beyond the immediate "sizing-up" of his
reading needs and interests, then recording that information is
unquestionably a needless expenditure of time and energy. If,
on the other hand, it is to become a permanent tool which en-

' Amy Winslow, "A Study of Data Pertinent to the Advising of Adult Readers."
Unpublished Master's thesis, Graduate Library School, University of Chicago, 1929,
Pp. 200.

10 Charles W. Mason, "Adult Education and the Public Library," Library Journal^
Vol. LVIII, No. 18 (October 15, 1933), pp. 830-32.

232. WHAT MAKES A BOOK READABLE

ables an adviser to serve a patron increasingly better, then in,
the long run, records seem both economical and practical.

In adult classes considerable information is usually obtained
from the student on his application for enrolment, but this is
probably used for administrative purposes more often than for
improvement of instructional techniques. The formalized char-
acter of the classroom tends to encourage the assignment of
reading to groups rather than to individuals. Hence, in classes
organized on the basis of ability, students may be given reading
that is fitted to their abilities rather than to their individual
interests and needs. Some investigation is necessary in adult
schools to determine what methods are being used in selecting
reading material, how satisfactory these methods are, and what
methods are most effective with different types of readers.

Turning from the reader to the book, we may ask, What qual-
ities about a book shall be considered in selecting one that will
be readable for a particular group of readers? This question
leads back to chapter ii, which gives a summary of opinion of li-
brarians, publishers, and others interested in adult education
with respect to what makes a book readable. The chapter also
indicates the nature of the problems that arise in defining a book
as readable or non-readable. They are the problems involved
also in selecting a readable book. For one cannot define a book as
readable save in terms of a particular group of readers. And
once it is defined in such terms, selection is relatively simple.

We need to know, therefore, the factors of content, style,
organization, and format which influence the readability of a
book for different readers who are identified with a particu-
lar class with respect to interests, education, and social and
economic status. Other investigators are endeavoring to discov-
er what people want to read about. From their findings, ques-
tions concerning interesting content are being answered. But
we do not know what methods of presentation, what stylistic
devices, what physical aspects of a book, or what forms of or-
ganization appeal to different sorts of readers. Each of these
and many other aspects must be defined for many classes of

HOW TO SELECT READING MATERIALS 233

readers before the selection of a readable book can be reduced
to a precise technique.

THE PRESENT APPROACH

In this chapter we have approached the problem of selecting
readable books from the point of view of inherent ease or diffi-
culty of content. We have taken the position that it is impor-
tant, among other things, that the book should be easy enough
to be readable. Not because ease is of necessity the primary
consideration in selecting a readable book that remains to be
discovered but because ease or difficulty is the aspect of read-
ability with which this study is chiefly concerned. And it is the
only aspect about which we have obtained objective informa-
tion. Again we repeat, by concentrating on difficulty we in no
way deny the importance of other factors in the total situation.
Unquestionably the reader's interest and zeal may tend to com-
pensate for inadequacies in reading ability and thereby alleviate
a part of the difficulty inherent in the material. These elements,
too, must be isolated for study. Ultimately all of them must be
considered in conjunction with ease or difficulty as determined
by structural elements in selecting a readable book for a particu-
lar reader.

Most persons who are engaged in guiding adult reading prob-
ably agree with Chancellor that

.... difficulty of text becomes a much less vital factor if the teacher or li-
brarian is able to tap a reader's keen desire to learn something about some
particular subject; that given this strong desire to get some facts and infor-
mation on his pet interest he will master unusual difficulties of vocabulary and
sentence structure in order to get the desired information, and that he will
learn to read incidentally."

This point of view, however, cannot wholly disregard difficul-
ty of material, if we are to accept scientific evidence concerning
the relationship between difficulty and enjoyment in reading.
For example, Burch reports that 80 per cent df the voluntary

" John Chancellor, "Available Reading Material for Native-Born Adult Illiterates
and Near-Illiterates." Mimeographed. Washington, D.C., Bureau of Prisons, Depart-
ment of Justice. 1933.

234 WHAT MAKES A BOOK READABLE

reading among junior and senior high-school pupils is chosen
from books which they read with an accuracy ranging from 60
to 90 per cent, with a median of 75 per cent. 12 One may infer
that material harder than this fails to produce in high-school pu-
pils the satisfaction and pleasure that are effected by an easy
book.

McAdams has compared the effect of permitting high-school
students to read only those books which they could comprehend
against the effect of allowing them to read any book. 13 Her find-
ings point to a marked relationship between enjoyment and
comprehension and between lack of enjoyment and difficulty.
She concludes that if high-school advisers will ascertain the level
of comprehension, attainable by each student and then suggest
books of an appropriate degree of difficulty, they may be able
to keep the student's reading interests alive and active.

A TYPICAL SITUATION

For the purpose of illustrating how the right book from the
point of view of structural difficulty may be selected for the
right person, let us assume a typical situation. An adult reader
comes to an adviser in library or classroom and requests a book
on adventure, a reading-list in economics, or, perhaps, some
other reading help. What the reader wants to read about is
therefore defined at the outset. His motive for reading is implied
in his request. Two other types of information are important.
The first answers the question: How well can the reader read?
The second is concerned with the query: What material bearing
on the desired subject is appropriate in difficulty to the reader '$
level of reading ability? Expressing the answer to both ques-
tions in common terms gives a simple basis for selecting books of
appropriate difficulty.

" Mary Crowell Burch, "Determination of a Content of the Course in Literature of
a Suitable Difficulty for Junior and Senior High School Students," Genetic Psychology
Monographs ', Vol. IV, Nos, 2 and 3 (August-September, 1928), p. 265.

** Mary Ann McAdams, "The Effect of Guidance on the Reading Interests of Tenth
Grade Pupils/' Unpublished Master's thesis, Department of Education, University of
Chicago, 1933, Pp. no.

HOW TO SELECT READING MATERIALS 235

The practicability and effectiveness of such a procedure for stu-
dents in high school has been demonstrated by McAdams in her
experimental study. As a first step in controlling pupil's read-
ing, she assigned to each student in two experimental groups
what she called an "enjoyment-score" in reading. This score
was based on comprehension as measured by three reading tests,
and was considered representative of the level of difficulty at
which a pupil may read with ease and pleasure. Three judges
then assigned to each of 230 books a c 'difficulty-score" obtained
by scaling the material through the use of Standardized Oral
Reading Paragraphs. The validity of this procedure had been
determined earlier by experimentation. 14 Pupils in the experi-
mental groups were requested to read only those books having a
difficulty-score equivalent to their enjoyment-score. Pupils in
the control groups were not assigned an enjoyment-score nor
were they restricted in their reading.

The findings of the study with respect to enjoyment and diffi-
culty have been given on page 234. They support the proposal
that reader scores representing ability and book scores repre-
senting difficulty can be expressed in common terms, and hence
can be used effectively in guiding book selection. Only 9,9 per
cent of one experimental group, and 26.3 per cent of the other,
rejected books on the ground of their being "too hard," whereas
70,5 per cent of the unguided group found books which they at-
tempted to read too hard. The results obtained by McAdams
show that although her technique does not entirely eliminate
unsatisfactory choices, it greatly reduces the chance of a pupil's
attempting to read a book which is too difficult to be readable.

Since the primary purpose of this chapter is to present possi-
ble plans whereby reading ability and reading difficulty may be
co-ordinated, it seems important to describe and to evaluate cer-
tain procedures which are used or which may be used to that
end. This is done in succeeding sections.

** Laura M. Larsen, "An Objective Method of Selecting Appropriate Geographical
Reading for Fourth-Grade Pupils." Unpublished Master's thesis, Department of Edu-
cation, University of Chicago, 1925. Pp. 73.

WHAT MAKES A BOOK READABLE

USE OF OBJECTIVE TESTS IN DETERMINING
HOW WELL ADULTS READ

The most scientific method of determining how well a person
reads is to measure his achievement by means of objective tests
which define ability in terms of standards usually expressed as
"grade norms/' In their study of children's reading, for exam-
ple, Vogel and Washburne used the Stanford Achievement Test
to determine reading scores and to establish the difficulty of
reading materials. 15 Dale and Tyler used the Monroe Standard-
ized Silent Reading Test in determining achievement of adults
of limited ability. 16 They then predicted the difficulty of techni-
cal material in terms of these scores. In the present study, rela-
tive ranks on the Adult Reading Test have been used to identify
"best" and "poorest" readers. Difficulty of general reading ma-
terial is estimated in chapter vi in terms of the average reading
score of "poorest" readers on this test.

Several advantages may be claimed for using reading tests,
In the first place, tests give a reliable index of reading ability in
terms of comparable scores based on actual performance. For
children, these numerical scores can readily be translated into
the grades to which they correspond. If a pupil makes a test
score equivalent to grade score 5.2, we know that he reads with
a degree of ability normally attained in the second month of
fifth grade. He may be expected, therefore, to read and compre-
hend most of the material written for fourth grade and the easi-
est material for fifth grade* A second advantage in measuring
reading ability by reading tests lies in the fact that test scores
may serve as a basis for grading books for difficulty. For exam-
ple, the formula devised by Vogel and Washburne for determin-
ing difficulty of children's books gives the reading score on the
paragraph-meaning section of the Stanford Achievement Test
necessary for reading the book measured. This score may be

** Carleton Washburne, The Right Book for the Right Child. New York: John Day
Co., 1933. Pp. 358.

* Edgar Dale and Ralph W. Tyler, "A Study of the Factors Influencing the Difficul-
ty of Reading Materials for Adults of Limited Reading Ability /* Library Quarterly, VoL
IV, No. 3 (July, 1934), pp. 384-4*^

HOW TO SELECT READING MATERIALS 237

translated into a reading grade which shows the correct grade-
placement for a particular book.

On the adult level the use of tests for measuring the ability of
the reader and for interpreting the difficulty of the material is
more restricted. The obstacles to be met in testing adults are
many and varied. Some are born of the testing idea itself, as
shown in chapter iii. Others are inherent in the adult reading
situation, which in general is not readily adaptable to a testing
program.

Adult reading is usually an individual, voluntary activity. A
person reads what he wants to read, provided that it is accessi-
ble and readable. Furthermore, he reads when and where it
pleases him to read. Only in organized classes is he one of a
group of readers whose reading is prescribed. Here the use of
tests as a part of the routine of instruction is both feasible and
desirable. Adult members of such classes co-operate willingly,
if they believe that the testing is related in any way to their edu-
cational advancement. For such readers there is need for read-
ing tests constructed from adult materials and accompanied by
norms for measuring successive stages of progress. Tests of this
sort are now under construction. It is hoped that by their use
one can obtain a measure of a reader's ability, which also will be
an index of the difficulty of materials that he can read under-
standingly.

Whether the use of tests for measuring reading ability is prac-
ticable in the library is another matter. Here adult-education
service is being organized systematically but less formally than
in the classroom. The librarian's advice or assistance is aimed
to guide those who are seeking self-instruction. "It must be of
first-quality, never superficial or haphazard. Neither must
there be anything pedagogic or official or superior about it/' 17
If advising becomes teaching, then there is danger that rapport
between reader and adviser will be disrupted and the seeker of
self-education will give up his enterprise offended or disgusted.

*? Sir Henry A. Miers, "Adult Education in Relation to Libraries," Library Journal,
Vol. LVIII, No. 8 (April 15, 1933), p. 339.

238 WHAT MAKES A BOOK READABLE

How far the librarian can go in measuring reading ability by
the use of tests without introducing a disastrous practice re-
mains to be discovered. Furthermore, from the standpoint of
time required to give and score tests, their use seems uneconom-
ical and impracticable in everyday library practice. Important
as it is that the librarian or readers' adviser should know the
reading ability of library patrons in order to give systematic,
sustained guidance, it seems probable that a less direct method
than testing should be followed at present in obtaining this in-
formation.

JUDGMENT OF READING ABILITY BASED ON KNOWLEDGE
OF MATERIALS PREVIOUSLY READ

Indirect evidence as to how well a person can read may be ob-
tained by the simpler and presumably more tactful method of
measuring reading ability, not in terms of his performance on a
test but in terms of the kind of material which he has read previ-
ously. This method is now widely used. As an adviser becomes
acquainted with a patron, she learns to know the type of mate-
rial he usually reads. If he asks for a book in a new field of inter-
est, she may be able to approximate the book he will find read-
able in that field by recalling or finding out the magazines or
books he has read.

This method has the advantage of being readily applied. But
it possesses serious limitations. In the first place one must rec-
ogni2e the fact that subjective estimates of human abilities im-
plied by human actions may or may not be reliable. The degree
of reliability depends on the validity of the action as an index of
the ability > on circumstances accompanying the action, and on
the discrimination of the person judging the ability. Although
these limitations are general, they have specific application for
the adviser who judges reading ability of a patron by his previ-
ous reading. Furthermore, the opinion of the adviser cannot be
translated into objective terms for another to use in selecting
reading material for the same patron.

As a result of the present investigation, it is possible to sup-
plement this general technique by using a classified list of ma-

HOW TO SELECT READING MATERIALS 239

terials representative of five areas of difficulty which imply a
similar number of levels of reading ability. The beginning of
such a list is to be found in chapter vi and in Appendix F. The
study of reading difficulty reported in that chapter needs to be
extended to include all books which appear best suited to read-
ers of average or below-average ability. With such a list, a li-
brarian should be able to define the difficulty of a book reported
read in terms which will indicate in a general way the area of
difficulty best suited to the reader's ability.

Obviously, the greater the amount of evidence obtained rela-
tive to previous reading, the greater the degree of accuracy with
which reading ability may be inferred. For example, if a person
reports having read and enjoyed Carroll, As the Earth Turns,
score .86; Colum, Cross-Roads in Ireland, score .94; and Over-
street, About Ourselves, score .91, it is fairly certain that he can
read understandingly books which we have ranked "easy."
This is not to say that he may not be able to read books ranked
"average" or "difficult." If, however, he reports that he cannot
understand Hemon, Maria Chapdelaine, score .3 1 ; Mason, The
Spell of Southern Shores, score .47; or Dimnet, The Art of Think-
ing, score .51 books whose respective themes are similar to
those of books read and enjoyed then the inference seems war-
ranted that these so-called "difficult" books are outside his
range of comprehension. Although we are assuming in this con-
nection that the rejection of a book as too difficult is the result
of structural handicaps, it is altogether probable that the diffi-
culty lies in part in other aspects of style or content. The libra-
rian who suspects such is the case may discover the source of
difficulty by questioning the reader or examining the particu-
lar book rejected. A more accurate estimate of the reader's abil-
ity will thus be obtained.

AN ESTIMATE OF READING ABILITY BASED ON INFORMA-
TION CONCERNING LAST YEAR IN SCHOOL

A third means by which some advisers estimate a patron's
reading ability, and hence the difficulty of materials which he
can read, is information concerning the amount of schooling he

WHAT MAKES A BOOK READABLE

has had or the last school grade attended. This type of informa-
tion has at least one advantage. It can be obtained easily. Two
distinct limitations attend the use of this information. In the
first place, the last year in school is not always a reliable index
of reading ability. Data reported in chapter iii indicate that
students at the secondary-school and college level have been
found to manifest a marked immaturity in their ability to read.
Some evidence has been reported which shows that the disparity
between educational progress and reading ability is considerably
less among new literates and adults whose formal schooling ter-
minated early.

In a study carried on in the opportunity schools of South Car-
olina, comparisons were drawn between reports of the last grade
attended in school and reading achievement as measured by the
Monroe Standardized Silent Reading Test. 18 Significant cor-
respondence was noted. For 226 adult students in four groups,
the average grade, according to statements made upon registra-
tion, and the achievement grade on tests were respectively: 2.3
and 2.2, 2.6 and 2.4, 5.3 and 4.9, and 6.4 and 6.9. When indi-
vidual records, rather than average records, were compared, it
was found that in only forty-nine cases was the discrepancy be-
tween years in school and ability on the test greater than one
grade. For those who claimed one or two years of schooling, the
score exceeded the claim by less than a year. The higher the
grade, the less the agreement between claim of previous school-
ing and reading-score, as shown in Table LVIII.

A close correspondence was found by Ojemann between read-
ing ability as measured by comprehension of parent-education
materials and number of years in school. 19 The subjects used
were 209 parents (mainly mothers) who were attending study
groups in a city in Iowa, the population of which approximates
30,000. Subjects having a partial or complete elementary-school

18 William S. Gray, Wil Lou Gray, and J. Warren Tilton, The Opportunity Schools of
South Carolina (New York: American Association for Adult Education, 1932), pp,
40-45.

x R. H. Ojemann, op. cit. y p, 32.

HOW TO SELECT READING MATERIALS

241

education made an average score on the Ojemann test of 18.98;
those having a partial or complete high-school education, 29 .25 ;
and those having more than a high-school education, 34.97. A
perfect score on the test is 45. Other things being equal, the
greater reading ability can be ascribed to longer educational
training.

Although we may assume that the agreement between years
of schooling and reading ability will be true generally, it appears
that for adults of little education more may be implied about

TABLE LVI1I

AVERAGE GBADE ACHIEVEMENT IN READING BY GROUPS

CLASSIFIED ACCORDING TO THE GRADE

REACHED IN PUBLIC SCHOOL

Number Cases

Grade Reached

Reading
Achievement

I .7

2,0

7.O

4,8

<.6

7, C

8 7

' J

reading ability than for those of broader educational back-
ground. In the present study, last grade in school, as reported by
756 adults, was correlated with reading ability as measured by
the two forms of the Adult Reading Test. Correlations of ,532
and -548 were obtained. While these correlations indicate that
some relationship exists between last grade attended in school
and reading ability, they show also that in individual cases
mere claims of attendance may not be a reliable index of reading
ability. Capacities and interests, experiences and opportunities
all may determine whether reading ability of an adult equals or
exceeds the normal expectancy of the last grade attended in
school.

242 WHAT MAKES A BOOK READABLE

A second limitation to the use of this method for estimating
reading ability lies in its interpretation. Let us assume that an
adult reader reports that he quit school in sixth grade. Let us
assume further that the reader's adviser rightly estimates his
reading ability at sixth grade. What does this mean in terms of
adult reading materials? Up to the present time there has been
no objective technique by means of which a reader's adviser
could determine what book is of appropriate difficulty for a
reader having a particular educational background and reading
ability.

Through the use of a classified list of materials this limitation
can be obviated and reading ability, as represented by years of
schooling, translated into concrete terms. For example, data
presented in the previous chapter lead us to presume that a per-
son whose attendance in school ended in third grade should find
"very easy" books simple enough to read with fair understand-
ing. These are the books having predicted scores greater than
1.15. According to the classification which we have adapted,
they belong to Area A. Persons reporting school attendance
ending at fourth or fifth grade will probably find "easy "books
best suited to their ability, provided, of course, that their edu-
cational pursuits since that time have not carried their reading
ability far above or below the normal expectancy of the lower in-
termediate grades. "Average" books, especially those ranking
at the lower extreme of Area C, may be read with ease and un-
derstanding by persons who claim a school attendance through
sixth grade. Books at the upper extreme of Area C presumably
represent a degree of difficulty that persons of junior high-school
training are able to read. Attendance through senior high school
or college implies ability to read "difficult" or "very difficult"
materials as they are defined in this study. That such an impli-
cation may assume a maturity of reading beyond the actual
truth must be granted in the light of evidence presented in chap-
ter in.

Probably the most to be claimed for this method is that infor-
mation concerning number of years in school furnishes a start-

HOW TO SELECT READING MATERIALS 243

ing-point for guidance. It locates the area of structural difficul-
ty at which a person presumably can read and the one beyond
which he probably cannot read easily. Greater precision in
defining a reader's level of ability depends on supplementary in-
formation relative to his reading interests and habits, obtained
by one or more of the several means suggested in this chapter.

ESTIMATING READING ABILITY BY THE EXPRESSED PREFER-
ENCE OF AN ADULT FOR PARTICULAR MATERIALS

A fourth method of determining the reading level of an adult
is based on the assumption that the expressed preference of a
reader for materials classified according to difficulty is a valid
index of materials best suited to his reading ability. According
to this method, an adviser should have available a variety of
materials representing different areas of difficulty. These mate-
rials are given to a new patron who comes for guidance, with the
instruction that he examine each and choose the one he prefers
to read* If he selects "easy" material in preference to that
which is "average" or "difficult/' it may be assumed that the
former is better suited to his needs and abilities. In making this
assumption, however, one needs to take into account a variety
of contributing factors inherent in the total situation. For ex-
ample, a reader may hesitate to choose very simple material,
particularly if it resembles children's reading. Hence, he delib-
erately overestimates his ability even at the expense of being
given too-difficult reading. Again, an easy book with attractive
format may cause a reader to select it in preference to others of
greater difficulty less pleasing in size and general appearance.
He accordingly defines his reading ability at a level that is
spuriously low.

In order to guard against the influence of format, sample
passages from books or magazines of varying degrees of difficul-
ty may be used. Examples of such passages are given here. Se-
lection A, an excerpt from the adapted Robinson Crusoe, repre-
sents "very easy reading, while Selection B, from People's
Popular Monthly > is representative of "easy" reading- Succeed-

244 WHAT MAKES A BOOK READABLE

ing selections, C and D, represent materials that are "average"
and "difficult," respectively.

It may be noted that "very difficult" materials classified at
Area E have been omitted. The reason for this lies in the fact
that the advisory situation with which we are dealing concerns
primarily the reader of limited ability. Presumably, he cannot
read materials at Area E. If a reader selects a passage from
Area D as the one he prefers to read, and if he is found actually
to read and enjoy "difficult" books, then it may be assumed
that he can read "very difficult" books at least reasonably well.
Should he not be able to read books of the latter sort, it seems
likely that other factors than those of structure influence the
difficulty of "very difficult" materials.

Despite the fact that the selected passages have been taken
from books whose predicted scores rank them in areas of diffi-
culty from A to D, respectively, their own predicted scores are
not identical with those for the total book. For example, the
predicted score for Selection A is 1*40, whereas the predicted
score for the entire book is 1,06. This discrepancy is to be ex-
pected, for the reason that the latter score represents the diffi-
culty of a whole series of passages from the same book, of which
Selection A is but a single sample. Both the sample and its
source, however, belong to the "very easy" area.

AREA A
(Scores 1.15-2,06)

ROBINSON CRUSOE'S BOAT

I needed a boat. So I cut down a great tree. It was five feet ten inches
across at the lower part, and four feet eleven inches at the top before it went
out into branches. I was twenty days cutting through it at the bottom, and
fourteen more days cutting away the branches. After this, it cost me a month
to make it into the shape of a boat outside, and three months more to cut out
the inside. Thus I made a very fine boat, big enough to carry me and all my
goods. When I had finished this work, I was very delighted with it. There
remained nothing but to get it into the water,

Defoe, Daniel, Robinson Crusoe. Adapted by Michael West. New York:
Longmans, Green & Co., 1931. Pp. 16-17.

Predicted score, 1.40

HOW TO SELECT READING MATERIALS 245

AREA B

(Scores .84-1.15)

FUN IN GARDENING

I like to put a bit of "kick" into my own vegetable gardening by experi-
menting, for I make my gardening "fun/* not work. One year I had a half
dozen kinds of tomatoes and about that many varieties of beans. I also like
to try new crops. A few years ago I successfully tried out Chinese cabbage,
an excellent salad crop, which likes the cool weather of fall or early spring.
When I hear of some new vegetable new to me or read of it in the many
seed catalogs I study each winter, I try to give this new fellow a few feet in
my garden, "just for the fun of it." People s Popular Monthly.

Predicted score, 1.02

AREA C

(Scores .53-^4)

HOW A GIRL SHOULD PREPARE HERSELF FOR A JOB

According to Harriet Houghton, director of vocational work in the Ameri-
can Woman's Association, the young woman should select her job as carefully
as she selects her husband. Unless it suits her it means mostly trouble. "Ten
years ago," says Miss Houghton, "the general-utility girl, the capable, all-
around, Jack-of-all-trades could make a place for herself; today it's the special-
ist everyone wants. The girl who would get ahead should decide the thing
she likes best to do and hew to that straight line. The intelligent way for a
girl to find the right thing is to look before she leaps analyze the job and
analyze herself and see how the two match up." Collier's.

Predicted score, .751

AREA D

(Scores .aa-,53)

THE CONTACT MAN IN WASHINGTON

The most conspicuous product of the bewildering multiplicity of bureaus
and overlapping of the so-called governmental functions of Washington today
is the neo-lobbyist, or Contact Man. Embodying the capacities of sight-seeing
guide, house detective, and automobile salesman, this comparatively recent
addition to the Capital's professional directory lives well by his ability to pene-
trate quickly the maze of red tape and petty bureaucratic formality surround-
ing every governmental official. He has at his finger tips such priceless, and
otherwise unobtainable, information as that the Weather Bureau is a function
of the Department of Agriculture, instead of the Department of the Interior.
He saves time and money for anyone who has business to transact with the
government. Atlantic Monthly.

Predicted score, .347

246 WHAT MAKES A BOOK READABLE

In these paragraphs length has been held relatively constant,
each paragraph containing approximately one hundred words.
Whether longer or shorter passages would promote more careful
discrimination between different degrees of difficulty can be de-
termined only by experimentation. Presumably, the passages
should be brief enough to prevent a needless expenditure of
time, yet long enough to insure a fairly reliable choice.

The futility of expecting to obtain a perfectly valid index of a
reader's ability by a single series of paragraphs must be admit-
ted at the outset. Inasmuch as the selected illustrations deal
with different themes, it is possible that a reader will choose Se-
lection B, let us say, because the content is what he wants to
read about rather than because it represents the most favorable
degree of difficulty. Hence, he overrates or underrates his usual
reading ability. In such circumstances, reading a second series
of paragraphs of the same sort may result in the reader's prefer-
ence for Selection C. Whether Area B or Area C represents the
difficulty of reading best suited to his ability is a question that
still needs to be answered. Presumably, the reading of several
series of paragraphs will finally give a reasonably satisfactory
index of the reader's ability. Such a procedure, however, is im-
practicable in the library where an adviser must determine a
reader's probable ability as expeditiously as possible.

Librarians have suggested that if the content of all the para-
graphs in a given series were held constant, and several series
prepared for use in particular situations, the technique described
here should bring effective and reliable information concerning a
reader's ability* Since each paragraph in such a series deals
with exactly the same topic as any other paragraph, a reader
will not be diverted from his purpose by the factor of content-
interest. Furthermore, the availability of several series, each
dealing with a particular theme, will mean that the patron seek-
ing guidance in reading biography, for example, reads a series of
paragraphs on biography and selects the one he enjoys most.
The reader of travel books will be asked to read sample passages
about travel, and so on.

HOW TO SELECT READING MATERIALS 247

Illustrations of the sorts of paragraphs which may be pre-
pared for this purpose follow. They deal with the common
theme, "The Truth about Africa/' yet each is representative of
a degree of structual difficulty different from any other in the
series. As in the, preceding illustration, this series includes but
four areas of difficulty. A, B, C, and D. Again, Area E is omit-
ted, for the reason that it represents a degree of difficulty pre-
sumably outside the realm of comprehension for limited readers.
The better readers who choose Selection D can probably read
most materials with a structural difficulty defined at Area E.

AREA A

Africa used to be called the dark continent. It was given this name be-
cause people knew very little about it. The name does not fit Africa today,
but many people do not know it. They still think of Africa as the dark conti-
nent of the world. They think that you cannot go into the center of Africa
unless you cut your own way through the forests. They think that people who
go to Central Africa are explorers. They do not know that people have been in
all parts of the continent, making roads and building towns. But that is just
what has been done. Now there are roads all over Africa, and business is
growing fast, We cannot call Africa a dark continent any longer.

Predicted score, 1.88

AREA B

People living in one country frequently have wrong ideas about another
country. One of our most popular ideas is that Africa is still a dark continent
as it was years ago before white men explored it. Many people believe that
the only way to enter the center of Africa is to blaze a trail through the
forests. So they look upon travellers from Central Africa as explorers. Ideas
like these could be quickly changed by a glance at a collection of modern
maps. They show that even the heart of darkest Africa has been fully ex-
plored and that everywhere the country is open to transportation. Africa is
too well-known and too far developed to be called a dark continent*

Predicted score, .917

AREA C

It is not uncommon for certain false ideas to be believed so persistently
that they become popular. A case in point is the popular belief that Africa is
still a dark continent with large areas quite unknown to white men* People
who cherish this idea believe that it is difficult to penetrate far into the interior
of the country, that one still blazes trails and visits places and people never
known before. They hail returning travelers from Central Africa as explorers.

248 WHAT MAKES A BOOK READABLE

Such ideas need not persist. One glance at a collection of modern colonial
maps would quickly establish their falsity. These maps show the heart of
darkest Africa as it is today fully explored, perfectly well-known, occupied,
and everywhere open to transportation.

Predicted score, .645

AREA D

Among the many popular delusions that are always with us there is none
more persistent than the current misapprehension about Africa. It still seems
to be believed that considerable areas of the once dark continent remain prac-
tically unknown to whites, that it is difficult to "penetrate" the remote in-
terior, that one still blazes trails and visits places and peoples hitherto un-
known. Returning travelers from Central Africa are hailed as explorers! Yet
this is a delusion which could easily be dispelled. One glance at a collection of
modern colonial maps would show the heart of darkest Africa as it is today a
rapidly developing commercial frontier perfectly well known, thoroughly
explored, occupied and everywhere open to transportation, Review of Reviews

Predicted score, .393

The selection representing Area D is an original passage taken
from an article in Review of Reviews, a magazine ranked in the
"difficult" class by the method described in chapter v. The pre-
dicted index of difficulty for the passage is .39. By rewriting the
content of the passage, with attention directed upon significant
elements, after the manner to be presented in the next chapter,
we obtained selections designated A, B, and C. Their indexes of
difficulty are respectively 1.88, .92, and .65.

In using series of paragraphs to discover reading ability, the
following procedure is suggested:

i. Find out the general field of interest of the reader, or the
specific topic in which he needs guidance.

a. Having discovered his field of interest, show him a series of
paragraphs dealing with that field. Ask him to read the various
selections and to indicate the paragraph he prefers to read.

3. If time permits, or if a reader is uncertain about a prefer*
ence, secure a second or a third choice by the use of other series
of selections on the same general topic.

4. Record the reading ability of the reader on his identifica-
tion card, in terms of Areas A, B, C, or D. This information be-

HOW TO SELECT READING MATERIALS 249

comes a permanent guide for later use in advising him about his
reading until such time as he requests harder materials.

The use of this method of determining the reading level of an
adult promises to have several advantages. From the point of
view of the reader there seems reason to believe that he will be
saved embarrassment or wounded pride if the adviser obtains his
preference in a perfectly frank but informal way. It seems prob-
able, too, that he will be better served by choosing the kind of
reading he finds most pleasant to read. From the point of view
of the library, it appears that an economy in time may be
effected by obtaining a measure of reading ability which simul-
taneously indentifies the appropriate difficulty of reading mate-
rials for the patron in question.

The reader who chooses a very easy paragraph defines his read-
ing ability in terms of materials at AreaA,any of which should be
readable for him, in so far as readability can be determined by
ease or difficulty of structural elements. But there is a marked
scarcity of material at this level, as shown by the responses to
Chancellor's inquiry, 20 by the testimony presented in Dicker-
son's report," and by the experimental findings in Gray's study
of opportunity schools in South Carolina. 23 If the reader chooses
Paragraph C, he claims ability to read materials that exist in
abundance. This is the area of the popular novel and magazine,
intended for the average or near-average reader. Should he
choose a selection at Area D, he identifies himself as a relatively
independent reader who can read whatever material interests
him and who finds in structural elements no very serious handi-
cap to understanding.

The accuracy of the reader's judgment concerning what he
can read most effectively depends upon several circumstances.
These include the approach made by the adviser, the reader's

w John Chancellor, op. cit.

L. L. Dickerson, Libraries and Adult Education, p. 58. Report of a study made by
the American Library Association. Chicago: American Library Association, 1926. Pp.
*8 4 .

William S. Gray, Wil Lou Gray, and J. W. Tilton, op, cit.

250 WHAT MAKES A BOOK READABLE

willingness to co-operate, and the reader's power to discrimi-
nate between the paragraphs in a series. Other data may be
used to check against the reader's judgment, such as last grade
attended in school, last book read, best-liked book, magazine
most enjoyed, and so on.

If we are to allow for the influence of other qualities than a
reader's ability, then we must admit that a self-definition of
ability may fail to be wholly reliable. For example, let us pre-
sume that a person is asked to read the second series of para-
graphs. Suppose, further, that he knows very little about Afri-
ca, which is the subject of the series. Paragraphs C and D seem
difficult, and he therefore chooses Paragraph B. Were another
series of paragraphs presented, the content of which is directly
related to his greatest reading interest, and about which he had
read widely, he might select Paragraph C, and thereby define his
ability quite differently and perhaps more accurately.

What we mean to emphasize is that information concerning
one's ability as it is determined by this method may vary with
interest in reading content, familiarity with the material, and
other qualities that influence what one is able to read. The more
closely the series corresponds to the field of the reader's interest,
the greater the accuracy that may be expected.

In presenting these methods of determining how well a reader
can read, we do not presume to say which is the best one to use.
This depends upon the advisory situation. We can go no farther
than to point out the advantages and limitations of each meth-
od. It remains for readers* advisers to put to practical test the
method, or combination of methods, best suited to their own
situations. To this end the selection of books may be made
more effective through a careful definition of each individual's
ability to read.

SELECTING READING MATERIALS OF
APPROPRIATE DIFFICULTY

When the adviser has learned how well a patron can read and
has interpreted his ability in terms of the difficulty of reading

HOW TO SELECT READING MATERIALS 251

materials, her next task is to select from available material that
which most nearly approaches the desired level. Were reading
materials distributed more evenly over the whole area of diffi-
culty, the task would be far less complicated. Of materials used
in the present study, relatively few were found to have an actual
or predicted difficulty equivalent to Area A. A somewhat larger
number rank at Areas B and D, and still more fall in the average
area, C.

What to give the reader who finds 'Very easy" material
adapted to his needs is a problem. It is perhaps more acute for
the teacher or adults of limited reading ability than for the li-
brarian. For the beginning adult reader frequently feels ineligi-
ble to library privileges. He reads what the classroom teacher
recommends or what he is able to find independently. Much of
the latter reading is not available in library or classroom, where
materials are chosen for their literary worth. It is available at
newsstand and cigar store in the form of wood-pulp magazines
and cheap fiction. In these "literary" dregs is to be found adult
content presented with a simplicity of expression which can be
approached by nothing else in the field of adult reading save
the few special adaptations shown in Figure 23. That many
persons do read this kind of material, either from necessity or
preference, is evidenced by the enormous circulation figures of
magazines of low-grade fiction. How to guide the reading in-
terests of such readers toward more wholesome satisfactions is
the problem with which adult workers and librarians constantly
are grappling.

But guidance in reading is far from being restricted solely to
persons of limited reading ability. In the library it is as compre-
hensive as the patronage requesting it, and that, as we have
said, represents a wide range of interests and abilities. Hence
librarians and readers' advisers daily find themselves directing
the reading activities of both good and poor readers. For,
whether by one of the methods just described, or by intuition,
librarians somehow do discover in most cases, perhaps who
are the good, the mediocre, and the poor readers. And having

252 WHAT MAKES A BOOK READABLE

made such a discovery they recommend or select books which
they believe will suit the needs of their patrons.

Undoubtedly much book-selection is made on the basis of
inspection, sometimes with the book actually in hand, frequent-
ly from memory. Often, a hasty thumbing through the card
catalogue or through book-lists, or a moment's reflective com-
parison of available books is enough, and the recommendation
is made. Again, the librarian-adviser engages in painstaking
search for an elusive article or book that promises to be the
right one for a particular reader. She has catalogued it mentally
as to content-appeal, difficulty, and other qualities which ex-
perience has shown are important in book selection.

Such practices as these are probably as old as the library it-
self. Often they result in the selection of the right book for a
particular reader the frequency of success probably being in
direct ratio to the insight and efficiency of the adviser. Some-
times they fail. And it is in the failure of personal opinion that
the need arises for a more objective means of selecting books
suited to the ability of adult readers.

As we outline possible procedures for classifying books with
respect to difficulty, we are assuming that such a classification
will aid librarians, readers' advisers, and teachers of adults in
selecting and recommending books for others to read. Whether
our assumption is sound remains to be proved by a comparative
study of book-selection as it is now carried on in specific situa-
tions and as it would be carried on through the aid of the more
objective methods proposed in the following sections.

In the advisory situation set up at the outset of this chapter
the reader's interest, it will be remembered, has been defined in
his request. The task of the adviser, therefore, is to select from
available material bearing on the reader's field of interest con-
tent suited to his reading ability. It is with this task in mind
that we present various methods of classifying reading materials
on the basis of structural difficulty. As was pointed out earlier,
other factors which determine difficulty must ultimately be con-
sidered.

HOW TO SELECT READING MATERIALS 253

CLASSIFYING READING MATERIAL BY A PREDICTED
INDEX OF DIFFICULTY

The most precise method now available involves the use of a
regression equation, as illustrated in chapter vi. By such a tech-
nique an index of difficulty is determined for all widely read
books or for all books that readers of limited ability presumably
can read. These indexes give a means of classifying books into
areas of difficulty ranging from 'Very easy" to "very difficult/'

Recording the difficulty-value for a given book inside the cov-
er or in the card catalogue gives a ready means of determining
whether the book is appropriate in difficulty for a reader of giv-
en ability. Probably the numerical index, as well as its relative
rank, should be recorded if the difficulty-value of a book is to be
of greatest service in an advisory situation. The reason for this
recommendation lies in the close relationship between difficulty-
values of books at extremes of adjacent areas. As pointed out
earlier, an average book (C) with a predicted score of .80, let us
say, is not significantly more difficult than an easy book (B),
with a predicted score of .86. For a reader of "B" ability, then,
one book is about as appropriate from the point of view of struc-
tural difficulty as the other.

Since the time required to predict the difficulty of a single
book may reach several hours, every effort should be made not
to duplicate computations. Doubtless the best and the most
economical service can be obtained through a central agency
which will calculate the index of difficulty for each book, inter-
pret this index in relative terms, and then transmit both types of
information to librarians and teachers of adults. Such an agen-
cy was instituted in connection with the present study under
financial sponsorship of the federal government. Its services
were used in rating the difficulty of the 350 books listed in
Appendix F.

This work represents only a beginning of what must be done
in simplifying and improving the adviser's task of selecting ma-
terials suited to a reader's ability. More of the old books must
be studied and their difficulty determined objectively. New

254 WHAT MAKES A BOOK READABLE

books need to be given an index of difficulty on accession. Pam-
phlets and brochures on a variety of topics should be rated for
structural difficulty and made available for ready recommenda-
tion in guiding adult reading. Finally, of course, classification
must be made on the basis of total readability for specific kinds
of readers when the necessary objective information for such a
classification will have been obtained.

RANKING READING MATERIALS FOR RELATIVE DIFFICULTY
WHEN ELEMENTS ARE CONSIDERED OF EQUAL WEIGHT

Many investigators have attempted to determine reading
difficulty by counting one or more significant structural ele-
ments inherent in the content. By this method, which has been
described in chapter v, all elements are assumed to bear some re-
lationship to difficulty. They are accordingly given equal
weight in ranking materials, whereas they, are given relative
weight when combined into a single agency of prediction repre-
sented by the regression equation.

A few investigators have gone farther and have suggested
tentative norms for different reading levels in terms of certain
elements. Lewerenz, for example, has created grade norms on
the basis of standardized tests and established a definite grade-
placement for percentage of occurrence of words beginning with
tv, h, b y /, or <?. 23 He has applied these norms extensively in de-
termining grade-placement of newspapers, textbooks, fiction,
and scientific books. 24 Since interpretation of the grade-place-
ment is in terms of comprehension as measured by the Stanford
Achievement Test, the method is more directly useful for grad-
ing children's reading than that of adults. Furthermore, the
findings of this study do not point to a significant relationship
between w, h> b, i, and e words and reading difficulty of adults,

Johnson has suggested tentative norms in percentage of poly-

3 Alfred S, Lewerenz, "Measurement of the Difficulty of Reading Materials," Edu
cational Research Bulletin (Los Angeles City Schools), VIII (March, 1919), 11-16.

8 4 Alfred S. Lewerenz, "Objective Measurement of Diverse Types of Reading Ma-
terials," Los Angeles Educational Research Bulletin^ IX (October, 1929), 8-11.

HOW TO SELECT READING MATERIALS 255

syllabic words in the different grades. 25 The use of this measure
gives results corresponding to the rating of books on the basis of
technical words, a method which was devised by Pressey. 26 Al-
though polysyllables have been found in the present study to
bear significant relationship to difficulty, other elements appear
to give a more reliable single index of difficulty for adults.

Estimating the level of difficulty to which a particular book
belongs, by use of elements of difficulty discovered in chapter iv,
involves two steps : first, counting the number or percentage of
certain significant elements as described in chapter v; and, sec-
ond, comparing the average occurrence of these elements with
the standard given in Table LXV, chapter viii, in order to desig-
nate the area of difficulty, A, B, C, D, or E. The standards shown
in this table give the range of occurrence and the median occur-
rence of five elements as well as the variation of occurrence in
terms of the upper and lower quartile. These measures were
found to characterize materials at successive areas of difficulty.
They were derived from the distribution of each element in
books whose average reading-score falls within the area repre-
sented. A total of 350 books listed in Appendix F were used in
obtaining these standards.

The steps to be followed in estimating difficulty by this meth-
or are:

1. Sample a book by selecting a series of passages approxi-
mately one hundred words in length, as before.

2. Analyze each passage for the occurrence of significant ele-
ments shown in Table LXV. The number of elements counted
depends upon the degree of precision desired. If, for example,
the use of three or four elements agrees with the standards of a
particular level in Table LXV, the results may be accepted as
adequate for all practical purposes.

3. Average the occurrence of the elements counted in the

*s George K. Johnson, "An Objective Method of Determining Reading Difficulty/'
Journal of Educational Research, XXI (April, 1930), 1283-87.

96 L. C. Pressey, "Determination of the Technical Vocabulary of School Subjects/'
School and Society XX (July 19, 1924), 91-96.

256 WHAT MAKES A BOOK READABLE

analysis to secure an average occurrence in an average hundred-
word selection.

4. Compare the obtained averages with the standards shown
in Table LXV.

5. Assign to the book the area of difficulty A, B, C, D, or E
which it most nearly represents. Agreement with the upper
quartile places the book at the difficult extreme of any level.
Agreement with the median places it midway at any level.
Agreement with the lower quartile ranks it at the easy end of
any leveL

Ranking by this method has been shown by limited investiga-
tion to be about as reliable as the use of a regression equation.
A coefficient of .54 was obtained by correlating the relative
ranking of several books of fiction with their absolute difficulty
expressed in terms of the average reading scores made by adults
of limited ability. A higher correlation of .66 was found between
estimated difficulty of non-fiction materials and their absolute
difficulty.

SCALING MATERIALS FOR DIFFICULTY

A third method of determining the difficulty of reading mate-
rial involves comparisons with specimens which constitute a
scale. For this purpose, the sample paragraphs representing
Areas A, B, C, and D, on pages 244-45, are taken as a scale of
difficulty. The difficulty of a particular book is the difficulty-
value of the specimen of the scale which it most nearly resem-
bles. This is determined by a technique of scale usage, which in
the field of tests and measurements is designated the "ascend-
ing-descending" method.

The method requires that a sample paragraph from a particu-
lar book be moved from A, the lowest step on the scale, toward
the higher steps until the judge decides that the specimen on the
scale is more difficult than the paragraph being measured. He
then begins with D, the highest point shown on the scale, and
compares the paragraph with successively lower steps until a
point is reached at which the specimen on the scale is easier than

HOW TO SELECT READING MATERIALS 257

the paragraph in hand. The paragraph then receives a rating
represented by agreement between the two rankings or by an
average of the two rankings. For example, if in ascending the
scale, a paragraph is ranked C, and in descending, it is ranked B,
its final value is either at the lowest extreme of C or the highest
extreme of B. The number of paragraphs sampled from a book
depends on the time available and the accuracy desired. One
paragraph from each chapter is generally adequate. An average
of the ranking of all the samples gives the index of difficulty for
the entire book.

Since experience in the use of scales for other purposes has
given evidence that two or more persons working together tend
to secure more satisfactory results than does the independent
worker, it is probable that more than one scaling is desirable.
Even then the results for the present use are probably less pre-
cise than those obtained by the regression equation or by rank-
ing on the basis of the occurrence of several elements of difficul-
ty. That the reliability of scaling tends to increase with training
and practice has been shown by previous investigation in other
fields. 27

In order to try out the foregoing method of scaling reading
materials for difficulty, we enlisted the co-operation of two li-
brarians interested in the plan. They were asked to scale twen-
ty-five books for difficulty by two different techniques. The
first involved the use of the Standard Oral Reading Para-
graphs, which were scaled on the basis of the rate and accuracy
of recognizing the words involved. The second involved the use
of specimens from the Adult Reading Test, which were scaled on
the basis of difficulty with respect to comprehension. In both
cases, ten paragraphs of each book were scaled by the ascending-
descending method, and the ten values thus obtained were aver-
aged to give a difficulty-value to the book as a whole.

The first method was based on pure inspection, that is, on the
visual matching of each sample paragraph with specimens on

*i W. S. Monroe, J. C. De Voss, and F. J. Kelly, Educational Tests and Measurements
(Boston*. Houghton MifHin Co., 1917), p. 158.

258 WHAT MAKES A BOOK READABLE

the scale. The second introduced analytical judgment of com-
prehension, that is, a comparison of the elements which influ-
ence difficulty of comprehension as they occur in the sample
paragraph and in the specimens on the scale. Both methods re-
sulted in notable agreement, the ratings tending to draw closer
and closer together as the scaling was continued. Conferences
between judges also tended to give uniformity to their ratings.

Both judges expressed a preference for the second method.
They agreed that an analytical study of elements of difficulty
accompanying the scaling process leads to greater accuracy than
pure inspection. A skilful reader, they contended, cannot make
comparisons easily without comprehending the meaning of the
passage. It is therefore a more familiar exercise than the visual
matching involved in sheer inspection.

Since the judges felt more assurance in using the second
method than the first, it actually took less time to analyze and
compare the paragraphs for difficulty of comprehension than to
scale them by general impression. It must be admitted that at
the outset both judges were somewhat skeptical of arriving at
any uniformity of rating by either method. At the termination
of several practice periods distributed through three or four
weeks, they expressed a confidence that the second method
would prove helpful in recommending books for adult readers
whose reading ability could be estimated.

Although it is important to discover that librarians agree on
the difficulty-value of a selection, it seems more important to
discover how reliable the value is on which they do agree. That
is to say, if they agree in their scaling, for example, that Cimar-
ron belongs at Area C, is their scaling correct? Or does Cimar-
ron belong at Area B ? It is at this point that the scaling method
of determining difficulty falls short of the two methods present-
ed in previous sections. We have calculated reliability coeffi-
cients to prove just how reliable will be the index of difficulty
obtained by use of the regression equation or by mere counting
of significant elements. The same degree of reliability, then,
may be expected for the difficulty-value of any book determined

HOW TO SELECT READING MATERIALS 259

by any person, granting his ability to do the simple arithmetical
processes necessary. Scaling, on the other hand, involves a cer-
tain degree of subjectivity. The extent to which this may be re-
duced determines, for the most part, the reliability of the values
thus obtained.

We have some evidence bearing on this procedure obtained
from the experiment just reported. Most of the twenty-five
books which the librarians scaled for difficulty had already re-
ceived an absolute difficulty- value from testing. It was possible,
therefore, to estimate the reliability of the values assigned by
the librarians. Not only did scaled values obtained by analysis
tend to agree more frequently with absolute values than did
those obtained by pure inspection, but variation from absolute
values tended to be less. It appears, then, that scaling can be
refined, if inspection is supplemented by comparing certain ele-
ments of difficulty in the sample paragraph with those in speci-
mens on the scale. As in earlier techniques, the elements used
for comparison include different hard words, personal pronouns,
average sentence-length, percentage of different words, and
prepositional phrases. Care should be taken to utilize para-
graphs of a length comparable to that of the specimens and to
express the occurrence of the elements as a ratio of one hundred
words.

Whatever the method used to classify materials for difficulty,
it must be not only economical of time for librarians and advis-
ers but reliable enough to prove a valuable aid in selecting the
right book for the right reader.

SUMMARY

This chapter has proposed several methods for the use of li-
brarians, readers' advisers, teachers of adults, and other persons
who are called upon to select materials of appropriate difficulty
for adult readers. Each method assumes that two types of in-
formation are essential One type is concerned with how well a
particular adult can read, the other, with how easy or difficult
is available material bearing on the field of the reader's interest.

260 WHAT MAKES A BOOK READABLE

Each method further assumes that these two types of informa-
tion will be most useful if they are translated into common
terms whereby a definition of reading ability automatically
identifies materials of appropriate difficulty.

Although the chapter has suggested techniques intended to
improve the selection of readable books, it does not claim to
solve the whole problem. It has, rather, reduced the total prob-
lem of selection by a consideration of one aspect, that of fitting
materials to the reader's ability when fitness is defined in terms
of appropriate structural difficulty. Beyond this, the chapter
has gone no farther than to indicate the kind of procedure need-
ed: first, in securing other information about a book than diffi-
culty; second, in reducing such information to a common de-
nominator of quality or quantity; and finally, in utilizing it in
the selection of a book a particular kind of reader will find read-
able.

How valuable these procedures may be in actual practice re-
mains to be determined. Each must be submitted to experi-
mentation in advisory situations of library and classroom. This
is necessary in order to discover whether the findings with re-
spect to structural difficulty actually improve the selection of
materials for particular readers. If evidence is found to prove
their practical importance, then there is promise that the identi-
fication of other aspects of readability may objectify and simpli-
fy the task of selecting appropriate reading materials for differ-
ent classes of readers.

CHAPTER VIII
HOW TO PREPARE READABLE MATERIALS

TTN TURNING from the problem of selecting readable books
I to the problem of preparing readable books, we are led to
-" ask a number of questions. What sorts of readable books
are now available? What qualities shall be considered in pre-
paring a book that is readable? What steps have been taken to
improve the readability of books for adults of limited educa-
tion? With what success have readable books been prepared for
such readers? How can information concerning elements of dif-
ficulty be used in making books readable for adults of limited
schooling?

Most of these questions pertain to the problem in general.
The last question points to one aspect of readability ease of
reading and asks how findings obtained by this study may be
used to improve the readability of books for adult readers in
general and for adults of limited education in particular.

So much has been said throughout this report of the need for
readable books that there is danger of being misunderstood.
We are not implying that there are no readable books. All of us
have read and are reading them. We have little trouble in label-
ing a book readable or unreadable for ourselves, despite the
lack of a thoroughgoing definition of a readable book. We are
less confident about readable books for others. As has been fre-
quently repeated, to label a book readable is to imply knowl-
edge of the reader, his interests, needs, tastes, and abilities. It
further implies knowledge of the book details of content, style,
format, organization all the aspects that may make a book
easy and pleasant for a particular person to read*

READABLE BOOKS ARE OF MANY KINDS

What are the books that we find readable? They are of many
kinds books of fiction and books on art, history, philosophy,

262 WHAT MAKES A BOOK READABLE

economics, geography, psychology, as well as other more or less
specialized subjects. Some readable books are intended for "in-
school" use in adult classes; many more are for general "out-of-
school/' voluntary reading. They tend to present fiction more
often than fact. Occasionally a readable book is also a "good"
book; more often it is, as Cheney has said in his survey of the
book industry, "one of hundreds which are good possibilities,
badly botched, or bubbles cleverly ballyhooed." 1

The reason seems clear. The good, readable book for out-of-
school reading has been too long the accidental product of a
writer who has the intuition to "sense" his reader's mind or who
can dissolve the jargon of a timely topic into language the man
of the street can understand. The good readable book for in-
school use has been the outcome of observations of particular
groups and analyses of what they need to read and what they
want to read. Few writers have found the secret of writing read-
ably for large groups of readers or, having found it, know that
they have it. A large number, on the other hand, fall short of
writing readably. This is true not because they lack the poten-
tialities for doing so. It is, rather, because, being engrossed
in ideas, they unconsciously neglect words and phrases which
make ideas easy and pleasant for the largest possible audience
to read. Their books may be "good," but they are not read-
able. These are the writers who presumably will profit most
from knowledge of what qualities make a book readable for a
particular class of readers.

TEW BOOKS ARE AVAILABLE FOR ADULTS
OF LIMITED ABILITY

Directors and teachers of organized classes for adults are vig-
orous in their contention that the educational progress of adults
of delayed schooling depends in a measure upon the preparation
of more material to meet their needs. These teachers and direc-
tors are demanding books of adult interest* They want brief

* 0. H* Cheney, Economic Survey of the Book Industry, *93<>"3i (New York: National
Association of Book Publishers, 1931), p* 98,

HOW TO PREPARE READABLE MATERIALS 263

books dealing with social and economic themes, presented in
simple form and style, and graded to conform with successive
stages of progress to the point where systematic instruction
ends.

Chancellor's inquiry shows that attempts to provide adults of
limited education with readable material for instructional use
have met with varying degrees of success. Textbooks written
particularly for use in Americanization classes commonly re-
ceive two sorts of criticism. One is directed against the patron-
izing or childish style in which the rudiments of information
bearing on American life are presented. The other criticism
points to the rapid increase in difficulty which makes the book
unreadable for the beginning reader of English. In some books
the difficulty progresses from primer level to the approximate
level of fourth grade, all within a range of two hundred pages or
less. There is little or no opportunity for the learner to gain flu-
ency at one level before proceeding to the next.

In some few instances these criticisms are obviated by grad-
ing the reading material within a series rather than within a
single book. But graded series of adult books are rare. Of the
few now available, the last book in the series tends to be least
satisfactory. It assumes a maturity of reading habits far beyond
what teachers of adults claim can reasonably be attained by the
use of the earlier books in the series. Slow progress in reading is
the inevitable result. With more adequate information concern-
ing the elements related to difficulty for adult readers, textbooks
can be graded more satisfactorily. For grading is a matter of
method. And method of how a thing is to be accomplished de-
pends upon what is to be accomplished. We may argue, there-
fore, that a definition of qualities which make a book readable
for certain classes of readers is the first step in preparing text-
books whose readability is attained not by chance but by design
and whose use will promote progress in reading for adult learn-
ers.

As already indicated, librarians are vigorous in their plea for
simple, out-of-school reading materials for adults of limited ex-

264 WHAT MAKES A BOOK READABLE

perience with books. What to select for men and women who
have done little or no serious reading, yet who suddenly show
interest in an informational field, is a perplexing problem. They
want books on Technocracy, Inflation, National Defense, War
Debts, the New Deal, and other timely topics. But what is
available on these subjects? Too often nothing more than pro-
found discussions of basic principles or abstract statistical trea-
tises. Frequently these are presented in an involved and ob-
scure style baffling to the reader whose understanding depends
upon concrete, human content, simply and lucidly written.

Objective evidence presented in Figure 23, chapter vi, sup-
ports the testimony of librarians and teachers of adult classes to
the effect that the supply of readable books for readers of lim-
ited ability is now inadequate. The figure indicates roughly the
relative amount of material available at Areas A, B, C, D, and
E. Within some limitations it shows that many books are read-
able at Area C, whereas the number at Area A is undoubtedly
limited. We do not mean to imply that difficulty is the pre-
dominant aspect of readability nor that a classification on that
basis presents the whole picture. Such a classification, however,
probably does more than hint at gaps in the long up-grade from
the adapted Robinson Crusoe to Jean Christophe,

Figure 23 also makes clear that many so-called "easy" books
must be made still easier if the adult who is just beginning to
read is to find material which will make possible his continuous
reading progress. What are commonly designated " better* *
books of biography, history, economics, and science need to be
written in a simplicity of language and brevity of statement
that are comprehensible to the reader of lesser ability who hope-
fully attempts to eke out limited education by voluntary read-
ing.

''BETTER'* THEMES BE PRESENTED SIMPLY?

A brief survey of experiments now being carried on gives
proof that "better" themes can be presented simply. In the
field of newspaper writing, we have American News which prints

HOW TO PREPARE READABLE MATERIALS 265

"the news of the week in Swenson's 900 words." Here we find
such so-called difficult subjects as "money plans/' "recovery
plans/' "preparation for war/' and so on, presented in a way
that reduces vocabulary difficulties to a minimum. The mate-
rial ranks in difficulty at Area B, as shown by a predicted score
of 1. 14. This score is obtained by use of five elements: different
hard words, personal pronouns, sentence-length in words, per-
centage of different words, and number of prepositional phrases.
Greater simplicity might be effected in this newspaper by de-
creasing the average length of sentences and reducing the num-
ber of prepositional phrases.

A second literary experiment is Ilin's New Russia's Primer,
which illustrates a recent attempt in Russia to enlighten mil-
lions of untutored persons concerning themes of great social sig-
nificance. Vital content, dramatic style, and simple expression
have been combined to produce a book readable for Russian
children from twelve to fourteen years of age. The English
translation by Counts and Lodge may be somewhat more diffi-
cult for the reason that the translators have aimed primarily to
preserve the spirit and substance of the original rather than its
simplicity. However, the difficulty of the translation is prob-
ably not greatly increased, as shown by the ranking of the Eng-
lish version at Area B in Figure 23, chapter vi. This is the area
which we have designated tentatively as appropriate for the
intermediate grades.

A third experiment worthy of mention is the simplification of
standard novels for adults who cannot read the originals. Not
all simplifications have been endorsed as readable for the class
of reader for whom they are intended. Frequently the adapted
book is an abridgment of the original which possesses qualities
of interest and literary worth but which lacks simplicity of ex-
pression. In some cases this abridgment is difficult for adults of
limited reading ability, for the reason that extent of vocabulary,
length and form of sentences, and other elements have not been
simplified enough to effect a satisfactory ease of reading. For
example, the predicted difficulty of Michael West's adapted

266

WHAT MAKES A BOOK READABLE

Treasure Island is approximately that of Stevenson's original
edition, when difficulty is measured by the occurrence of certain
structural elements. The index of difficulty predicted for the
former is .69; for the latter, .64. The chief difference between
the two versions, as shown in Table LIX, is in the number of
different hard words. This difference is the result of the author's
intention to bring the original novel within the vocabulary of

TABLE LIX

AVERAGE OCCURRENCE OF FIVE ELEMENTS OF DIFFICULTY IN

HUNDRED-WORD SAMPLINGS OF TwO VERSIONS

OF Treasure Island

Original

Adapted

(Predicted

Elements of Difficulty

(Predicted

Score, 0.64)

Score 0,69)

20.2

Number of different hard words

13-0

9.92

Number of personal pronouns

10.4

a8. 3

Average length of sentence in words

29.2

68.0

Percentage of different words

69.5

11,7

Number of prepositional phrases

9.0

the Ntw Method Readers lA-V. This vocabulary contains 1,779
words, of which 1,669 are am ong the 2,000 commonest words in
the English language. 2

A second illustration of the relatively slight modification
made in structural difficulty may be found in the adaptation of
Moby Dick y by Sylvia Chatfield Bates. The predicted difficulty
of the original is o.a8, which ranks the book at Area D. The
adapted text has a predicted score of 0.75 which gives it a rank-
ing at Area C. The most marked difference between the two
books is found in length. Melville's Moby Dick is a long novel of
some 500 to 600 pages. Bates has told the gist of the story in
about 115 pages.

Differences in the occurrence of structural elements may be
seen in Table LX. It presents the average occurrence of each

9 Michael West, New Method Readers for Teaching English to Foreign Children,
Descriptive Booklet. London: Longmans, Green & Co., 1929* Pp. 76.

HOW TO PREPARE READABLE MATERIALS 267

element in a series of hundred-word samples, one from each
chapter of the two books. But three elements appear to be
markedly changed by simplification: percentage of monosylla-
bles, percentage of simple sentences, and average sentence-
length in words. A relatively high percentage of different words
persists in the simplified text despite the author's effort to re-
duce the range of vocabulary through the use of two word-lists,

TABLE LX

AVERAGE OCCURRENCE OP SEVEN ELEMENTS OF DIFFICULTY IN

HUNDRED-WORD SAMPLINGS OF TWO VERSIONS

OF Moby Dick

Original
(Predicted
Score, 0.28)

Elements of Difficulty-

Adapted
(Predicted
Score, 0.75)

29.1
42.0

73-5
27.0
32.2
8-7
"3

Number of different hard words
Percentage of monosyllables
Percentage of different words
Percentage of simple sentences
Average length of sentences in words
Number of personal pronouns
Number of prepositional phrases

22.2

75-4

70.9
51.0

18.2

9.8

10.3

Rej all's 4,000- Word List and the first 5,000 words in Thorn-
dike's List. The range of selection possible from these two lists
obviously does not insure a highly restricted vocabulary.

What changes have been made in adapting other classics for
readers of meager reading experience ? Tables LXI-LXI V an-
swer this question for four books: Robinson Crusoe, Silas Mar-
ner, Les Miserable*, and The Vicar o/Wakefield. Predicted scores
for the first three original texts, 0.26, 0*13, and 0.20 respec-
tively, place them at Area E with respect to structural difficulty.
This is the area of 'Very difficult" books. The predicted scores
for the same three adapted texts, 2.06, 1.24, and 1,26, rank them
at Area A. They are "very easy" books, representative of the
simplest material now available for adults. A considerably less
degree of simplification is manifested by West's adaptation of
7 he Vicar of Wakefield. Its predicted score, .82, marks it an

268

WHAT MAKES A BOOK READABLE

"easy" book as against a score of .25 for the "difficult" original
version. It is evident from the tables that the difficulty of the
original Robinson Crusoe lies in length and structure of its sen-

TABLE LXI

AVERAGE OCCURRENCE or SEVEN ELEMENTS OF DIFFICULTY IN

HUNDRED- WORD SAMPLINGS OF Two VERSIONS

OF Robinson Crusoe

Original
(Predicted
Score, 0.26)

Elements of Difficulty

Adapted
(Predicted
Score, 2,06)

19-3
76.9
67.0
4,0

74-5
12,3

n ,i

6.0

3-3

41.4
32.0

12.0

'3-5
9.8

TABLE LXII

AVERAGE OCCURRENCE OF SEVEN ELEMENTS OP DIFFICULTY IN

HUNDRED- WORD SAMPLINGS OF Two VERSIONS

OF Silas Marner

Original

Adapted

(Predicted

Elements of Difficulty

(Predicted

Score, 0.13)

Score, 1,24}

28.3

Number of different hard words

i 4 .6

72.0

Percentage of monosyllables

75.8

71.9

Percentage of different words

65.5

14.0
41.7

Percentage of simple sentences
Average length of sentences in words

46.0

13.0

9-3

Number of personal pronouns

3-4

12,0

Number of prepositional phrases

6.7

tences. These two elements have been notably simplified* Silas
Marner and Les Miserables illustrate what an author can accom-
plish by modifying several elements in the direction of simplicity
of expression-

HOW TO PREPARE READABLE MATERIALS 269

A fourth experiment the writing of scientific and other types
of material in basic English represents an effort to present
better themes simply. It will be commented upon later in its
relation to the Basic English Vocabulary List.

TABLE LXIII

AVERAGE OCCURRENCE OF FIVE ELEMENTS OF DIFFICULTY IN

HUNDRED- WORD SAMPLINGS OF Two VERSIONS

OF Les Mis erases

Original

Adapted

(Predicted

Elements of Difficulty

(Predicted

Score, 0.20)

Score, 1.26)

29.9

Number of different hard words

8.0

5-7

Number of personal pronouns

13.0

41.2

Average length of sentence in words

10.9

68.3

Percentage of different words

66.4

13-1

Number of prepositional phrases

8-5

TABLE LXIV

AVERAGE OCCURRENCE OF FIVE ELEMENTS OF DIFFICULTY IN

HUNDRED- WORD SAMPLINGS OF Two VERSIONS OF

The Vicar of Waksficld

Original
(Predicted
Score, 0.25)

Elements of Difficulty

Adapted
(Predicted
Score, 0.82)

25.8
10.3
35-6
73-o

12.8

Number of different hard words
Number of personal pronouns
Average length of sentence in words
Percentage of different words
Number of prepositional phrases

20. o

13-9

iS.8

70.2
10.3

WHAT SHALL BE CONSIDERED IN PREPARING
A READABLE BOOK?

The four experiments just mentioned seem to point to a defi-
nite tendency among a few authors to write down to the under-
educated millions who lack the reading experience of the larger
proportion of adults. What is the secret of writing readably?

270 WHAT MAKES A BOOK READABLE

Is it some kind of a magic art? Or is it a technique acquired
through persistent effort to create that which a particular au-
dience will find easy and pleasant to read? For example, do
Will Rogers and Arthur Brisbane possess a peculiar gift of se-
lecting topics of high social, civic, and personal interest and of
presenting them simply and vitally? Are these two writers con-
sciously aiming to reach the masses rather than the more dis-
criminative reader? Whatever the answer, there is no denying
that they are writing what the reader of Dream World can read
and understand and what the reader of Atlantic Monthly may
read if he chooses.

Evidence in support of this statement is shown by Lewerenz,
who has secured interesting facts relative to the degree of sim-
plicity characterizing certain types of newspaper material. He
has graded the material for vocabulary-difficulty by the use of
a grade-placement formula described earlier. 3 His findings show
that in Will Rogers' "Remarks" there is found a vocabulary
diversity adapted to a reading grade-level of 4.6. According to
Lewerenz this is about the level represented by the ordinary
comic strip. Brisbane's news comment 'Today" contains a di-
versity of vocabulary which is equivalent to grade 6.2. This is
the level generally maintained by editorials in the newspapers
read most widely by the working-class. It is also the level of
local and sporting news a level at least two years below the
editorials found in newspapers most commonly read by profes-
sional and business groups.

If we examine the writings of Rogers and Brisbane we dis-
cover other qualities than clear words and brief sentences that
presumably are related to readability. These writers are inter-
preters of the news. Their field is the universe and all that is in
it. They present their subjects briefly, directly, and vividly.
Their style is simple and unpretentious; their tone, informal and
personal How much each of these qualities contributes to mak-
ing their writing readable, we cannot say* For one reader sim-

a Alfred S. Lewerenz, "Vocabulary Grade-Placement of Typical Newspaper Con-
tent/' Lot dngelts Educational Research Bulletin, X (September, 19,30), 4-6.

HOW TO PREPARE READABLE MATERIALS 271

plicity of expression may be paramount in making the material
readable. For another, simplicity may be of less consequence
than the concreteness of the story. For still another, it may
be the subtle force of these writers that makes what they write
readable. There is apparently need for research concerning com-
munication of thought through print as there is concerning
communication by radio or by any other medium, to discover
the qualities most useful in reaching different classes of audi-
ences.

If, for example, the format of a book adds to readability, we
need to know further the size of book, the color and quality of
its binding, the number of pages, and other features most de-
sired by different classes of readers. Undoubtedly the format of
Les Miserable*, Silas Marner, and Robinson Crusoe (adapted)
makes them far more readable for adults of limited reading
ability than they were originally.

Writers and publishers need to inquire further concerning the
size of type and the kind of illustration preferred by different
readers. If organization contributes to readability, how do dif-
ferent classes of readers want a book organized? Are paragraph
headings desirable at Area B, let us say, and not at Area D ? Do
footnotes confuse the reader of materials at Area A, but not the
more able reader of materials at Area D ? From the point of view
of content, what theme is of greatest interest to different read-
ers? Is the general topic better suited to one class than to an-
other? Should a theme be defined more narrowing for readers
of books at Area A? What is the nature of the concepts that
can be comprehended by readers who are limited to Areas A
and B?

In regard to style, the writer should ask such questions as:
What class of reader finds realism pleasing and phantasy baf-
fling? Is narration more readable at one level than is description ?
How far does a personal, informal style improve readability for
particular readers? What structural elements interfere with ease
of reading? What kind of vocabulary is most readable at a par-
ticular level ? To what degree are length and form of sentence

272 WHAT MAKES A BOOK READABLE

related to ease,, and hence to readability for different classes of
readers ?

Some of these questions and others of a similar sort have been
investigated in the field of children's reading with the result that
materials for children show marked improvement in readability.
If they can be answered with the same or with a greater degree
of definiteness in the field of adult reading, it seems reasonable
to anticipate two major outcomes: one, the formulation of prin-
ciples of readable writing for use in preparing books for differ-
ent kinds of adult readers; and the other, the ultimate develop-
ment of standards by means of which existing books may be
judged readable.

WHAT CONTRIBUTIONS HAVE SCIENTIFIC STUDIES

MADE TO THE PROBLEM OF PREPARING

READABLE MATERIALS?

For many years the fact has been recognized that books for
children must be adapted to successive reading levels. Publish-
ers of juvenile books, especially textbooks, have been striving
to create educational and recreational material suited to the
needs and interests of children of various ages. Previous to the
development of scientific techniques in 1910, it was difficult to
determine objectively when a book was appropriate for children
at a particular age- or grade-level. Subjective opinion dominat-
ed. Publishers and authors largely felt their way along*

Gradually objective evidence has accumulated with respect
to the reading attainment which may be regarded as normal at
each grade-level, the vocabulary usage and other language hab-
its of children of different ages, and the reading interests and
preferences which characterize successive stages of child de-
velopment* Concerning reading material, evidence is now avail-
able relative to several fundamental aspects. Among these are
the color, size, and general appearance of a book which children
of certain ages find most attractive; the kind of illustrations
which appeals to them; the length and kind of sentence and the
extent of vocabulary best fitted to their successive stages of

HOW TO PREPARE READABLE MATERIALS 273

progress; and the size of type which they find most readable at
different grade levels.

With these data at hand, writers and publishers are producing
increasingly more readable textbooks and supplementary books
for children's use in the classroom. Much is still to be desired
in making the general trade book readable for children at dif-
ferent stages of reading progress*

Considerably less has been accomplished in the grading of
reading material for adults. The reasons are many. The most
important, perhaps, is that there has been a certain apathy
toward adult reading. If a person reached maturity without
having learned to read, the fact was recognized, but practically
nothing was done about it. If he could read but didn't read, it
was because he "never cared for reading" or "never had time to
read," or "never could find anything he wanted to read." These
pseudo reasons settled the matter.

During the past few years public attitude has changed. We
have begun to examine the character of adult reading and to
inq\iire concerning the facts related to it. Organized attempts
at adult education and self-directed efforts toward the same end
have opened up new opportunities for the scientific study of the
subject. Barriers to investigation increasingly diminish as we
become accustomed to fact-finding techniques. Most of us will-
ingly admit our reading tastes and preferences and the nature
of our reading habits, for we know that the information we give
will be lost in the composite responses of thousands of other
persons.

Some of the facts now available pertain to the reading inter-
ests and habits of adults. Others relate to adult reading materi-
al to its vocabulary "burden" and to other elements now rec-
ognized as indexes of ease or difficulty. Still others present a
graphic picture of the problems and processes of book publica-
tion. It is through the utilization of all these outcomes of scien-
tific research that writers and publishers will be able to prepare
a greater number of readable books for different classes of read-
ers. A brief summary of important studies and their implica-

274 WHAT MAKES A BOOK READABLE

tions for writers and publishers is given in the sections that
follow.

WHAT HAS BEEN DISCOVERED ABOUT THE READING
INTERESTS AND HABITS OF ADULTS?

Reference has been made earlier to the initial report by Gray
and Monroe on the reading interests and habits of adults. 4 This
study has contributed to an understanding of such topics as the
status of reading in American life, the amount and character of
material read, the interests and motives of adult readers, the
influences that affect the development of reading interests and
motives, and the importance of establishing permanent reading
habits early. Inasmuch as these facts are broad and general,
their chief value to publishers is to emphasize that more analyt-
ical studies must be made if books are to serve the varied needs
and purposes of a wide range of readers.

One aspect of the entire problem of adult reading reported by
Gray and Monroe pertains to interests of adult readers. Ob-
viously, what people want to read about is something which
publishers should know. Much information is now provided by
Waples and Tyler, who have made a detailed study of the
problem. 5 They have compiled group scores to be used in iden-
tifying the groups most interested in a given subject of non-
fiction*

Suppose a publisher plans to produce a book on foreign trade.
What groups may he expect to interest? The evidence from
Waples' study shows that prisoners, farmers, factory girls, and
college students will have little or no interest in it. High-school
teachers, telephone operators, commercial students, postal
clerks, and others will evidence average interest in it. But no
group will be highly interested. 6 This does not mean that these
groups will not read the book, for Waples has discovered that

* William S. Gray and Ruth Monroe, The Reading Interest* and ff obits of Adults.
New York: Macmillan Co., 1929. Pp. 305*

* Douglas Waples and Ralph W. Tyler, What People Want to Read About* Chicago;
University of Chicago Press, 1931. Pp. 31 a.

HOW TO PREPARE READABLE MATERIALS 275

the correlation between what one wants to read and what one
actually reads, although positive, is not significant. 7 What the
evidence does mean is that publishers cannot expect the same
degree of interest in a book on foreign trade as in one on prices
and cost of living, let us say. This latter topic, according to
Waples and Tyler, is one of more general interest for adult read-
ing. Hence, it stands a greater chance of being read.

VOCABULARY NEEDS AND READABLE WRITING

Notable contributions to the intricate problem of readability
are found in standard vocabulary-lists which are the products
of scientific investigation. The Thorndike list of 10,000 words,
published in 1921, was the first contribution of value to writers
and publishers in determining the frequency and importance of
words in printed material. 8

In compiling this list and his later one of 20,000 words, 9
Thorndike has made two assumptions. The first is that his ma-
terials from which words are selected are representative. The
second is that the most frequent words are the most important.
Although these assumptions have been frequently challenged,
much evidence has been found to support them. For example,
the most frequent thousand words in Thorndike's list have a
high frequency of occurrence in all types of material, both
literary and scientific. 10 The conclusion seems warranted that
adult reading necessitates familiarity with at least the first
thousand words on his list, but that writers and publishers can
presumably place less confidence in the importance of words
classified beyond the first thousand.

There is a tendency to compile basic vocabulary lists for the

i Douglas Waples, "The Relation of Subject Interests to Actual Reading/' Library
Quarterly, II (January, 1932), 42-70.

8 E. L. Thorndike, The Teachers 1 Word Book. New York: Teachers College, Colum-
bia University, 1921.

9 E. L. Thorndike, The Teachers' Word Book of 20,000 Words. New York: Teachers
College, Columbia University, 1931.

10 Edgar Dale, "Evaluating Thorndike's Word List," Educational Research Bulletin
(Ohio State University), X (November 25, 1931), 451-57.

276 WHAT MAKES A BOOK READABLE

purpose of modifying the vocabulary difficulties of adults who
are either learning or who have just learned to read. Some of
these lists have special application to reading materials for
foreign-born adults.

Cornell has prepared a tentative list of 2,27 words as a begin-
ning reading vocabulary for such persons, with special reference
to those who cannot read in any language." The list is composed
of 164 words common to five textbooks for adults and to the
first 500 words either of the Thorndike or the Gates list;" 25
words in the first 100 of Thorndike and Gates not found in the
five texts; 16 words that seem necessary to complete concepts,
even though they do not appear in either of the first two classes;
and 22 street signs. Cornell intended that her list be used not
only to provide a basic vocabulary for beginning reading but
also to furnish a basis for classification of adults for early in-
struction in reading. In preparing first lessons in reading for
foreign-born adults learning to read English, the list seems to
promise valuable help.

Swenson's more recent list of 300 English words has been pre-
pared with the view of formulating a minimum vocabulary for
foreigners learning to speak English. 13 According to the findings
of the Language Research Committee, who co-operated in its
compilation, the 300 words are all a foreigner needs in asking
for the things necessary for existence. Frequently this list is
used also in teaching foreigners to read. In such instances, its
adequacy is open to question* Three hundred words can hardly
give an understanding of the printed information found in an
English-speaking community. A longer list seems essential to
carry the mere beginning of English usage to mastery in either
speaking or writing. Such a list known as "Swenson's 900

Ethel Cornell, A Beginning Reading Vocabulary for Foreign-born Adults with Special
Reference to Those Who Cannot Read in Any Language. University of the State of New
York Bulletin, No, 948 (Albany, New York: University of the State of New York,
*93)> P- 24,

M Arthur L Gates, A Reading Vocabulary for the Primary Grades, New York: Teach-
ers College, Columbia University, 1926, Pp. 24.

Elaine Swenson, "Swenson's 300 Words/* New York Times> November 5, 1933.

HOW TO PREPARE READABLE MATERIALS 277

Words" is being used in a project described earlier, which aims
to print the news of the week so that beginning readers may
read it understandingly.

A more extensive word-list than either of the preceding two
has been prepared by Rejall. 14 This list includes 4,000 words
which every citizen and voter should know. The complete vo-
cabulary is divided into an elementary list of 1,000 words, an
intermediate list of 1,500 words, and an advanced list of 1,500
words. These sub-lists are primarily for the use of teachers in
selecting the essential words to be taught in progressive order.
Word recognition in silent reading rather than oral pronuncia-
tion is the basis for testing a foreign student's knowledge of
words in the list. Writers of textbooks for use in Americaniza-
tion classes frequently aim to utilize the 4,000 words in this
list.

From his work among Oriental students of English, Faucett
has come to regard about 1,500 words as the minimum vocabu-
lary for reading and understanding. 15 They are the "wide range"
words which comprise almost all the form-words needed in
normal and modern English prose. Stated in another way, they
are the indispensable and essential words which together make
up over 75 per cent of the word-occurrences in normal English
and constitute its great linguistic framework. Beyond this
framework are the "narrow-range" words, that is, the useful and
special words whose word-value make them neither indispensa-
ble nor essential.

This word-list is intended, according to its compilers, to be
"useful to those interested in establishing the minimum vocabu-
lary to be incorporated in all series of readers, in fixing a graded
vocabulary scale for supplementary readers, in furnishing ex-
aminers and school inspectors with a measuring rod, in helping
teachers and students to develop a sense of word-values, and in

*4 Alfred E. Rejall, "Reading Vocabularies," Thirty and One Reading Tests Jor Voters
and Citizenship (New York: Noble & Noble, 1926), pp. 49-69.

** Lawrence Faucett and Itsu Maki, A Study of English Word-Values Statistically
Determined from the Latest Extensive Word-Counts, Tokyo, Japan; Matsumura San-
shodo, 1932. Pp. 252.

278 WHAT MAKES A BOOK READABLE

making possible standardized tests." 16 Although an arbitrary
fixing of the limit of narrow-range words may inflict an injustice
on writers of textbooks and of general trade books for new
literates, yet the minimum use of such words and the more ex-
tensive use of wide-range words would seemingly produce a
more readable book than is possible by indifference to word-
values,

Ogden's Basic English vocabulary aims to meet the interna-
tional need at two different stages, 17 One is the stage of ordi-
nary communication which Ogden claims may be achieved in
idiomatic English through the use of 850 basic words and 124
international words. The other is the more advanced stage of
scientific internationalism, achieved by means of the 850 basic
words, 300 international words, an additional TOO words for
general science, and 50 more for any particular science. A great
economy is claimed in learning a number of phrases composed
of the same few words arranged in a different order over learning
a number of different words. For this reason it appears that the
foreign-born adult will be less handicapped by difficulties of
vocabulary in reading basic material than in reading other types
of materials provided that he has mastered the basic words.
Actually, however, such is not the case. As critics have pointed
out, there are "stretchings" and shifts of meaning which a read-
er must acquire if he is to understand basic materials. 18 And
these stretchings involve the same or about the same amount of
learning effort as new words*

The few translations into basic English now available have
been made for two purposes. One is to furnish practice for the
learner of the new language. With this purpose we arc not
greatly concerned here. The other is to experiment in the writ-

*/Mf.,p. 8.

w C K, Qgden> Basic English. London: Kegan Paul, Trench, Trobner & Co.> 1931.
Pp. 96.

11 Michael P. West, E* Swenson, and Others, A Critical Examination of Basle JKn^
lish, Bulletin No* a of the Department of Educational Research* Ontario College of Edu-
cation, University of Toronto. Toronto*, University of Toronto Presa, 1934* Pp, $3,

HOW TO PREPARE READABLE MATERIALS 279

ing of books in an abbreviated language. 19 No claims are made
for literary excellence of translations into basic English beyond
what the originator of the language ascribes "to the rigid econ-
omy of words, which may be held by some to present a whole-
some antidote to the prevalent verbosity/* How much the prac-
tice of so rigid an economy, if such there is, may destroy the
pleasure of a book and make it less readable has not yet been
ascertained.

Since the 850 basic words are designed to furnish the widest
possible range of substitutes for other words, together with a
minimum inventory of common objects, they arc not neces-
sarily short words, nor are they "easy" words as that term has
been defined in the present study. Furthermore, inasmuch as
translators of basic English are concerned primarily with the
usage of a simplified language, their use of other structural ele-
ments related to difficulty or ease of reading is apparently more
a matter of chance than intention. Table LXXX V, on page 343,
shows that Carl and Anna ranks in difficulty at Area B. This is
the area of difficulty of reading materials which persons of
fourth-, fifth-, and sixth-grade reading ability normally can read
with ease and understanding. Other basic translations, however,
have been found to rank at higher areas of difficulty. It appears,
therefore, that although basic English as a literary medium
may produce easy reading, the level of difficulty is not the same
for all translations.

An examination of the various vocabulary-lists just described
shows that they do not agree. Some are short. Others are long,
In one case the list 5s derived from an analysis of general read-
ing. In another it is compiled from an analysis of such practical
sources as political and historical documents and papers, liter-
ary requirements of certain states, and reports from teachers
of adult classes concerning vocabulary needs of their pupils* In
some cases> words common to two lists arc designed to meet

'* C. K, Ogdcn, "Basic English * an International Language," Ntw Era, Vol. XIV,
No, i (Jftnu*ry

WHAT MAKES A BOOK READABLE

both conversational and reading needs. Others are intended pri-
marily to provide a basic reading vocabulary.

Which list or lists, then, is a writer or a publisher to use? And
how shall he use them? The answer to the first question is de-
termined by the audience he means to serve. Is it the foreign-
born adult who cannot read in any language? Or is it the for-
eigner who can read in his native language and who is learning
to read English? Or, again, is it the one who is learning to read
and speak English simultaneously? Is it the native-born adult
who is just learning to read? Or is it the reader who is improving
his reading ability? Is it the Oriental student of English? For
all these types of reading audience, the first few hundred words
in the Thorndike list are useful, since they are common to all
reading needs. For a specific reading audience, suitable words
may be determined by reference to other lists. It is probable
that any carefully prepared list represents a considerable ad-
vance over the judgment of an individual author or publisher
regarding the importance of a word.

With respect to the second question, it seems reasonable to
believe that material will be generally more readable for a par-
ticular group if the vocabulary is guided by the use of the most
suitable list. Teachers of adults generally take the position that
if basic lists were supplemented by words common to the adult
experiences of particular groups, their usefulness would be ex-
tended. Further research is needed to determine what words
represent adult motives, attitudes, and activities common to
different racial and sectional, occupational and cultural groups?.
When adult word-lists of this sort are prepared, we shall prob-
ably re-define "easy" and "hard" words on the basis of their
familiarity for adults as measured by adult experience*

Vocabulary-lists, then, represent an advance step in making
books more readable for adults of limited reading experience.
But vocabulary diversity 5s only one element related to diffi-
culty of reading, and hence to readability. And attention to one
element is not enough. We may restrict our writing vocabulary
and still produce material that presents obstacles to the reader.
The words may be difficult because they are long or unfamiliar*

HOW TO PREPARE READABLE MATERIALS 281

The difficulty may lie in the use of long and involved sentences,
frequent prepositional phrases, figures of speech, and other in-
dexes of difficulty. It may be influenced by an absence of easy
words, personal pronouns, short and simple sentences, and other
indicators of ease. Or, again, it may lie entirely outside the
realm of structural difficulty in some distracting aspect of con-
tent, format, style, or organization.

That significant differences in difficulty may exist among ma-
terials having a common vocabulary is shown by the selections
which follow. These paragraphs have been adapted from Swiss
Family Robinson by Michael West, whose simplifications prob-
ably arc known to all teachers of adults. The writing of these
paragraphs was motivated by an examination of the findings of
this study presented in earlier chapters. After such an examina-
tion, West generously volunteered to prove that something more
than a limited vocabulary is necessary to produce very simple
writing, and hence very easy reading.

In each selection West has restricted the vocabulary to 133
different easy words. The total number of words in the three
selections is approximately identical- 293, a88, and 293 words,
respectively. A comparison of the predicted indexes of diffi-
culty shows that although all selections may be classified as
"very easy/ 1 Selection B is significantly more difficult than Se-
lection A, and Selection C more difficult than Selection B. Had
the vocabulary been held constant, yet been made to include
harder words, the same variations in difficulty would have clas-
sified the selections into different areas of difficulty* While at-
tention to vocabulary, then, may produce easy reading, greater
case is attained by taking other significant elements into ac-
count.

I LOVE LOBSTER

SELECTION A

(Predicted score a. 10)

The two tubs were in the water. The tubs were near our boat. I wanted to
get them onto the land. I tried* I found that I could not do it. The bank waa
too steep. I could not get the tubs up the steep bank. So I set out to find a
better place* Jutt aa I set out I heard a cry* Jack wa crying out for help*

28o WHAT MAKES A BOOK READABLE

both conversational and reading needs. Others are intended pri-
marily to provide a basic reading vocabulary.

Which list or lists, then, is a writer or a publisher to use? And
how shall he use them? The answer to the first question is de-
termined by the audience he means to serve. Is it the foreign-
born adult who cannot read in any language? Or is it the for-
eigner who can read in his native language and who is learning
to read English? Or, again, is it the one who is learning to read
and speak English simultaneously? Is it the native-born adult
who is just learning to read ? Or is it the reader who is improving
his reading ability? Is it the Oriental student of English? For
all these types of reading audience, the first few hundred words
in the Thorndike list are useful, since they are common to all
reading needs. For a specific reading audience, suitable words
may be determined by reference to other lists. It is probable
that any carefully prepared list represents a considerable ad-
vance over the judgment of an individual author or publisher
regarding the importance of a word.

With respect to the second question, it seems reasonable to
believe that material will be generally more readable for a par-
ticular group if the vocabulary is guided by the use of the most
suitable list. Teachers of adults generally take the position that
if basic lists were supplemented by words common to the adult
experiences of particular groups, their usefulness would be ex-
tended. Further research is needed to determine what words
represent adult motives, attitudes, and activities common to
different racial and sectional, occupational and cultural groups.
When adult word-lists of this sort are prepared, we shall prob-
ably re-define "easy" and "hard" words on the basis of their
familiarity for adults as measured by adult experience.

Vocabulary-lists, then, represent an advance step in making
books more readable for adults of limited reading experience.
But vocabulary diversity is only one element related to diffi-
culty of reading, and hence to readability. And attention to one
element is not enough. We may restrict our writing vocabulary
and still produce material that presents obstacles to the reader.
The words may be difficult because they are long or unfamiliar.

HOW TO PREPARE READABLE MATERIALS 281

The difficulty may lie in the use of long and involved sentences,
frequent prepositional phrases, figures of speech, and other in-
dexes of difficulty. It may be influenced by an absence of easy
words, personal pronouns, short and simple sentences, and other
indicators of ease. Or 5 again, it may lie entirely outside the
realm of structural difficulty in some distracting aspect of con-
tent, format, style, or organization.

That significant differences in difficulty may exist among ma-
terials having a common vocabulary is shown by the selections
which follow. These paragraphs have been adapted from Swiss
Family Robinson by Michael West, whose simplifications prob-
ably are known to all teachers of adults. The writing of these
paragraphs was motivated by an examination of the findings of
this study presented in earlier chapters. After such an examina-
tion, West generously volunteered to prove that something more
than a limited vocabulary is necessary to produce very simple
writing, and hence very easy reading.

In each selection West has restricted the vocabulary to 133
different easy words. The total number of words in the three
selections is approximately identical 293, 288, and 293 words,
respectively. A comparison of the predicted indexes of diffi-
culty shows that although all selections may be classified as
"very easy," Selection B is significantly more difficult than Se-
lection A, and Selection C more difficult than Selection B. Had
the vocabulary been held constant, yet been made to include
harder words, the same variations in difficulty would have clas-
sified the selections into different areas of difficulty. While at-
tention to vocabulary, then, may produce easy reading, greater
ease is attained by taking other significant elements into ac-
count.

I LOVE LOBSTER

SELECTION A

(Predicted score 2.10)

The two tubs were in the water. The tubs were near our boat. I wanted to
get them onto the land. I tried- I found that I could not do it. The bank was
too steep. I could not get the tubs up the steep bank. So I set out to find a
better place. Just as I set out I heard a cry. Jack was crying out for help.

282 WHAT MAKES A BOOK READABLE

"Help!" "Help!" he cried. He might be in some danger! He was not far
away from me. I took my axe. I ran towards him. Jack was standing in a
deep pool. He was crying out in fear. When I came nearer I saw why he was
crying out. A big lobster had caught hold of his leg. He was very frightened
at it. He kicked. The lobster held on. He kicked. The lobster still held. I
went into the water. I held the lobster by the back. I pulled. The lobster let
go. We brought the lobster to land. Jack was soon quite happy again. He
said, "Let us take the lobster to mother." He caught the lobster in his hands.
But the lobster hit him. It hit him with its tail. He threw the lobster down
on the ground because he was angry. He hit the lobster with a big stone. I
became sad. I said, "You are acting like a foolish little child. This lobster has
done harm to you. So it is your enemy. So you have hit it. You want to do
harm to it. This is a show of anger. Do not be angry with your enemies. Do
not do harm to those who do harm to you. You should do good to them.
Love your enemies,"

SELECTION B

(Predicted score 1.93)

The two tubs were in the water near our boat. I wanted to get them onto
the land. I tried to do this, but could not, for the bank was too steep. I could
not get the tubs to land so I set out to find a better place that was not too
steep. Just as I set out, I heard Jack cry out for help. He might be in
some danger! I took my axe and ran towards him. He was not far away
from me. I found him standing in a deep pool of water crying out in fear.
When I came nearer to him I saw that a big lobster had caught hold of his
leg. He was very frightened. He kicked and kicked but the lobster still held
on. I went into the water, and took the lobster by its back, and pulled. It let
go, and we brought it to land. Jack was soon quite happy again and wanted
to take the lobster to his mother. ''Let us take it to mother," he said. He
caught it in his hands, and it hit him with its tail. He threw it down on the
ground in anger and hit it with a big stone. I became sad at this show of
anger. "You are acting like a foolish little child, Jack," said I. "You are
angry with the lobster and have hit it, and want to do harm to it. Do not be
angry with your enemies and try to harm them. Do good to those who do
harm to you. You should love your enemies."

SELECTION C

(Predicted score 1.69)

I wanted to get the two tubs that were in the water near our boat onto the
land; but when I tried to do this I found that the bank was too steep so that I
could not get the tubs to land, but had to set out to find a better place. Just as
I set out to find it I heard a cry for help from Jack who was not far away, and,
taking my axe, ran towards him fearing he might be in some danger. As I
went I saw that he was standing in a deep pool, and, as I came nearer, I saw
why he was crying out to me in fear. A big lobster had caught hold of his leg.
Being very frightened he kicked, but kick as he might the lobster still held on.

HOW TO PREPARE READABLE MATERIALS 283

Running into the water, and taking the lobster by its back, I pulled it away.
When it let go we brought it to land. As soon as Jack became quite happy
again, he said that he wanted to take the lobster to show to his mother, but,
when he caught it in his hands, he was hit by the angry lobster's tail, and be-
came so angry that he threw the lobster down on the ground and hit it with a
big stone. Being sad at this show of anger, "Jack," said I, "you are acting
like a foolish little child. Being angry with the lobster because it has done
harm to you, you hit it so as to do harm to it. I have said to you, have I not,
that we should do good to those who do harm to us and should not harm
them ? We should love our enemies."

ELEMENTS OF DIFFICULTY AND THEIR CONTRIBUTION
TO READABLE WRITING

If, as Cheney has said, difficult books, after school as well as
in school, are a prime unmaker of readers largely because
"good" books are too complex for the reading level of the largest
audience, then it seems reasonable to assume that readers would
not be "unmade" could they find good material written in a
manner they could understand. And we may further assume
that such material will be written when we have scientific infor-
mation regarding what qualities make a book easy or difficult.
Information of this sort is being rapidly extended to include
other elements of expression than vocabulary.

In Russia psychologists have been working on the problem
of discovering a medium of communication that will reach the
uneducated peasantry. For example, they have found that as a
means of mass influence the "humoristic" safety poster that is
characteristic of America is better understood and has greater
influence than the "bloody" type predominating in Russia. 20
Consequently, the posters now being produced are adopting the
former aspect. Psychologists have taken stenograms in meet-
ings of peasants, workers, and soldiers and compared them with
stenograms of speeches of educated persons. They find that vo-
cabulary and grammatical structure of the sentences differ ac-
cording to the training of the speaker. The more educated he
is, the greater the percentage of nouns in his speech and the
smaller the frequency of verbs. Uneducated persons use sen-

ao Ninth International Congress of Psychology Proceedings and Papers (Princeton,
New Jersey: Psychological Review Co., 1929), pp. 404-6.

284 WHAT MAKES A BOOK READABLE

tences that are longer but poorer in thought quality. These
findings have led to experimentation in increasing the intelligi-
bility of the juridic laws for peasants. From 40 to 90 per cent
increase has been secured merely by substituting verbs for many
nouns and by expressing the laws in sentences which contain
relatively few new ideas.

Reference has been made in an earlier chapter to the study of
Dale and Tyler concerning the elements of difficulty in techni-
cal reading matter. They have limited their study to one topic,
personal health, which is known to hold high interest for adults.
Their findings show that three elements are the best indicators
of the difficulty which adults of limited reading ability will meet
in reading material of this nature. These elements are: number
of different technical words, number of different hard, non-
technical words, and number of indeterminate clauses. By re-
ducing the occurrence of these elements in a given selection an
author will make the material more comprehensible for readers
of lesser ability.

In the present study we have assumed that to make a book
easy to read from the point of view of structural elements is a
long step in the direction of making it readable. We have ac-
cordingly identified elements that are related to difficulty for
the general adult reader and for the reader of lesser ability. The
exact relationship of these elements to difficulty has been pre-
sented in chapter iv.

For the latter group, the reader of lesser ability, the following
elements in reading material indicate difficulty.

1. Number of words not known to 90 per cent of sixth-grade pupils

2. Number of different hard words

3. Average sentence-length in syllables

4. Average sentence-length in words

5. Percentage of different words

6. Percentage of polysyllables

7. Percentage of different words not known to 90 per cent of sixth-grade
pupils

8. Maximum syllabic sentence-length

9. Minimum syllabic sentence-length
10. Number of figures of speech

HOW TO PREPARE READABLE MATERIALS 285

11. Number of prepositional phrases

12. Number of different words

13. Number of / words

14. Range of syllabic sentence -length

15. Number of infinitive and prepositional phrases

1 6. Percentage of content words

17. Number of asides

1 8. Percentage of bisyllables

19. Number of clauses introduced by subordinate conjunctions

20. Number of e words

21. Percentage of complex sentences

22. Number of b words

23. Number of clauses introduced by conjunctive adverbs

24. Percentage of compound and compound-complex sentences

The order of arrangement of the foregoing elements indicates
their worth as indexes of difficulty. That is to say, the number
of words in a book not known to 90 per cent of sixth-grade pupils
is a better index of its difficulty than the percentage of com-
pound and compound-complex sentences which it contains. The
first fifteen elements are the only indicators of difficulty that are
significant from a statistical point of view. We are hardly justi-
fied, however, in ignoring the other elements. The very fact
that they are related to difficulty at all is evidence that complex
sentences, asides, and bisyllables, for example, are indexes of
difficulty of material containing them.

If it is advantageous to know the elements which indicate
difficulty, it is equally advantageous to know the ones indicative
of ease. Such elements are:

1. Percentage of easy words

2. Number of easy words

3. Percentage of monosyllables

4. Number of personal pronouns

5. Number of explicit sentences

6. Number of sentences per paragraph

7. Number of third-person pronouns

8. Number of infinitive phrases

9. Number of simple sentences

10. Number of complex sentences

11. Number of clauses introduced by relative pronouns

12. Percentage of structural words

13. Number of h words

14. Number of compound-complex sentences

286 WHAT MAKES A BOOK READABLE

1 5. Number of w words

1 6. Number of first-person pronouns

17. Percentage of simple sentences

1 8. Number of compound and compound-complex sentences

19. Total number of words per paragraph

20. Percentage of compound-complex sentences

What do these lists of elements mean to the writer or the
publisher? In a general way we may say that an increase in the
occurrence of elements in the first list increases the difficulty of
a selection, whereas an increase of elements in the second list
decreases the difficulty. In other words, if we look for indexes
of ease in the adapted Robinson Crusoe, we find that its sen-
tences are shorter, simpler, and more explicit; and its words
shorter, easier, more familiar, and less diversified than in the
original text.

The author, therefore, who wants to reach the widest possible
audience will write simply. He will plan how he can express his
ideas so as to reduce complexity. But will such planning tend
to make a book of poorer quality? Will it reduce the freshness
and spontaneity of the author's style? Will it kill the art of
writing? If it does, we may ask further, need it do so? Does not
the author plan his plot? Do not the actions, words, even the
thoughts of his characters follow a pattern ? Is not the organiza-
tion of a book so planned that chapter divisions mark shifts of
time and place? Even the most carefully planned book may be
the most artistic and the most spontaneous.

To suggest, then, that writers plan to utilize words and
phrases that will be understood by the greatest possible audi-
ence is merely to add one more aspect to the planning which
they do anyway. The hope of the author who writes for a re-
stricted audience lies in knowing the standards of that audience
in terms of needs and interests. Such standards have not yet
been developed and perhaps never can be for all aspects of read-
ability.

For those aspects that are objectively measurable, however,
there seems reason to believe that fairly reliable standards can
be developed. A beginning has been made in this direction for

HOW TO PREPARE READABLE MATERIALS 287

structural elements related to difficulty, as shown in Table
LXV. The tentative standards presented in this table are given
in terms of the range of occurrence, the median occurrence, and
variation in occurrence of five elements in 350 books classified
at Areas A, B, C, D, and E. They were derived by distributing
the average occurrence of each element in all books classified at
each of the various levels. Data in the table are read as follows:
the average number of different hard words per 100 ranges from
21.4 to 6.0 for "very easy" books; from 25.5 to 12.0 for "easy"
books; from 33.2 to 12.0 for "average" books; from 36.2 to
20.6 for "difficult" books; and from 37.0 to 19.3 for "very diffi-
cult" books. The median percentage of different hard words
characterizing the five areas of difficulty is 12.3, 20.3, 24.3,
28,63 and 31.1. Three-fourths of all material at Areas A, B,
C, D, and E contain a percentage of different hard words be-
low 16.6, 21.9, 26.5, 30.9, and 33.7, respectively, while one-
fourth contains a percentage below 8.7, 16.9, 22.1, 26.5, and
28.2. Variations in the occurrence of other elements are read
similarly.

Although variations in the occurrence of different elements
are not always sharply defined, they do indicate definite trends,
which are shown graphically by Figures 27-31. The median for
each element shows a relatively consistent gradation from one
area to the next. There is an increase in number of different
hard words, in length of sentence, in number of different words,
and in number of prepositional phrases; and a decrease in num-
ber of personal pronouns. The upper quartile of one area closely
approximates the lower quartile of the next higher area, indi-
cating that materials at one extreme of a particular area are not
significantly different in structural difficulty from materials at
the opposite extreme of another area. Such a circumstance is in
accord with gradations in reading ability as shown by standard
tests. An examination of reading norms for succeeding grades
shows that although those for fifth and sixth grades, let us say,
are markedly different, the norm for high fifth and low sixth
grades differ almost imperceptibly.

288

WHAT MAKES A BOOK READABLE

TABLE LXV

TENTATIVE STANDARDS FOR USE IN WRITING FOR ADULTS
OF LIMITED READING ABILITY

AREAS OF DIFFICULTY

Very
Easy
1.15 to
2.06
(10 cases)

Easy
.84 to
1.15
(70 cases)

Average
53 to

.84
(i 8 8 cases)

Difficult

.22 tO

.53

(69 cases)

Very
Difficult
-.26

tO .22

(13 cases)

Number of different hard words
in a hundred-word paragraph
Range

21.4-6.0
16.6
12.3

3.7

16.8-1.9

13-5
11.5
9.2

24-^-9.3

20.8

14.8

12.0

64.6-36.5

59-7
54.6
51.6

H.3-5.8
10.7

8.4
6. 7

25.5-12.0
21.9
20.3
16.9

15.3-1.2
ii. 6
8.6
4-i

27.1-13.3

21. 1
ip.l
17.7

7I.I-53.8
6 7 . 7
65.7
61.5

14.6-8,6
12. 1

10,8

9,6

33.2-12
26.5
24-3

22.1

15.5-2.0
IO.4
8.2

5-8

38.5-13.0
25.8

22.8
20.4

74.7-58.9
7O.2
68.9
67.5

15.2-8.4
I2. 4

n.8
10.9

36.2-20.6

30-9
28.6
26.5

11.0-3.9
8.9
7.0
5-3

44.4-20.1
31.8
28.5
25.8

75.7-65.0
72.9
70.8
69.4

15.8-10,1

13-5
12.7
12.5

37-0-I9-3
33-7
31-1

28.2

12.3-2.6

8-5
5,0
3-8

74-5-34-1
42.8
40.0
35-5

73.9-67.0

7*-3
71.7
68,1

15,8-11.1
'5-5
i3>7
13*

Upper quartile

Median

Lower quartile

Number of first-, second-, and
third-person pronouns in a
hundred-word paragraph
Range

Upper quartile

Median

Lower quartile

Average sentence-length in
words in a hundred-word par-
agraph
Range

Upper quartile

Median

Lower quartile

Percentage of different words m
a hundred- word paragraph
Range

Upper Quartile

Median

Lower quartile

Number of prepositional phrases
in a hundred-word paragraph
Range ,

Upper quartile
Median ,

Lower quartile

HOW TO PREPARE READABLE MATERIALS 289

What does Table LXV mean for the writer of adult material?
In the first place, it gives him a measuring device by means of

Number
40

U.L.

Md.

5 L.L.

Area A Area B Area C Area D Area E

FIG. 27. Occurrence of different hard words in reading materials at successive
areas of difficulty.

Number
20

U.L.

Md.

L.L,

Area A Area B Area C Area D Area E

Fio, 28. Occurrence of personal pronouns in reading materials at successive areas
of difficulty.

which he can estimate the difficulty of his writing from the
point of view of structure without use of the regression equation.

Area A Area B Area C Area D Area E

FIG. 29. Average sentence-length in reading materials at successive areas of diffi-
culty.

Per Cent
80

75
70

65 U.L,
60 &

55 Md.
50 ft

L.L,

Area A Area B Area C Area D Area E

FIG. 30. Occurrence of different words in reading materials at successive areas of
difficulty.

HOW TO PREPARE READABLE MATERIALS 291

For example, if in several hundred-word samplings of his ma-
terial he finds that the occurrence of various structural elements
approximates their occurrence at Area C, then he has sufficient
evidence for assuming that his writing presents no more than
average difficulty, that it ranks with other average books shown
in Figure 23.

In the second place, Table LXV offers tentative standards
which a writer can use as guides in reaching a particular audi-

Number

Area A Area B Area C Area D Area E

FIG. 31. Occurrence of prepositional phrases in reading materials at successive
areas of difficulty.

ence. Let us suppose that he wishes to prepare material for
adults who are just beginning to read that is, material which
will rank at Area A. According to the standards in the table, he
should use on the average not more than about 8-1 6 different
hard words per hundred, 51-60 different words, and 6~io prepo-
sitional phrases. He should make a presentation so direct that
the number of personal pronouns will not be fewer than 9-13
per hundred words. Furthermore, his average length of sentence
should not exceed ia-ao words.

How shall a writer set about to meet these standards? He
may begin by reading materials classified at Area A. They will
include children's textbooks for grades a and 3, the simplest of
simplified classics shown in Figure 23, cheap wood-pulp maga-

292 WHAT MAKES A BOOK READABLE

zines materials that will give him a concrete notion of the
degree of simplicity represented at Area A. When he has caught
the "feel" of very easy reading, he can then begin to write,
pausing occasionally to check his work against desired stand-
ards and to make certain that he is imposing no structural diffi-
culties on the beginning reader.

Obviously, this is not the whole task of writing readably even
for the one class of reader that we have been considering, name-
ly, the reader of limited ability. There still remain many other
qualities whose relationship to readability needs to be deter-
mined for this type of reader. These have been suggested from
time to time throughout the present report. For other classes
of readers to write simply may mean making a book unreadable.
Perhaps it is not simple writing and hence ease in reading that
makes a book readable for them, but a combination of other
qualities. What those qualities are needs to be discovered also.
Again, it must be said that each quality should be isolated in
turn, studied as objectively as possible, and the findings be or-
ganized in a way that will aid writers and publishers in provid-
ing readable materials for all classes of readers.

APPENDIX A

The following list was used to discover the trend of opinion among libra-
rians and others concerning What Makes a Book Readable. A discussion of
the findings obtained by the use of this list appears in chapter ii.

AN INVESTIGATION TO DETERMINE FACTORS
OF READABILITY IN BOOKS

EXPLANATION OF LIST

The attached list includes possible factors of readability in books. It has
been compiled from letters received from librarians, readers' advisers, pub-
lishers, and other persons interested in adult education in response to an in-
quiry concerning the factors which in their judgment contribute to the read-
ability of books. The list is organized into four major categories: (I) Format
or Mechanical Features; (II) General Features of Organization; (III) Style of
Expression and Presentation; and (IV) Content. Each major category is di-
vided into sub-items, designated by Arabic numerals; and each sub-item is ex-
plained or qualified by a number of factors, suggested by the correspondents
as being possible factors affecting readability. These are designated a y b, c,
etc.

USE OF LIST

The list is sent to you for use in helping to evaluate the relative influence of
each category and each sub-item on readability; and in determining what fac-
tors are important under each sub-item. This step is preliminary to a more
objective study of their significance.

METHOD OF SCORING AND CHECKING

Three columns are set up at the right of each page, designated A, B, C.
Column A is to be used for checking specific factors; Column B, for evaluating
sub-items; and Column C, for evaluating the four major categories.

First step. Examine the complete list of factors and note the sub-items
and categories into which they have been classified, in order to familiarize
yourself with the general set-up of the list. You will note that there is direct
contradiction among some factors, apparent overlapping of others, and close
similarity among others. This has resulted from including all factors of possi-
ble readability, suggested by the correspondents.

Second step. Look over the factors, designated a,b, ,etc., and decide which
ones, in your judgment, make for readability. Indicate your decision by
checking such factors (V ) in Column A. If you wish to show that certain fac-
tors are of special significance, indicate by (V V). Leave blank spaces after
factors that you believe are insignificant or do not make for readability. Space

294 WHAT MAKES A BOOK READABLE

has been left for additional factors. Please include any that you think have
been omitted.

Third step. Consider the total value of the sub-items, I, 2, 3, etc., in each
major category as equal to 100 points. Look over the sub-items in Category I,
and evaluate their relative importance in promoting readability. Distribute
the total value, loo points, among these sub-items to indicate their proportion-
ate values. Write the values in Column B, opposite each sub-item. Be sure
that the sum of all values assigned to sub-items i, a, 3, etc., equals 100 points,
which is the total value of Category I.

Next 3 do the same thing for Category II; then for Category III, and finally
for Category IV. In each case, distribute 100 points among their respective
sub-items, writing in Column B. Each category should total 100 points.

Fourth step. Now consider the total value of all four Categories, I, II, III,
IV as equal to 100 points. Decide what proportion of 100 points best repre-
sents the value of each category in influencing readability. Distribute the 100
points among the four categories, as your judgment dictates. Write the as-
signed value in Column C after each category. The sum of the four values
should total 100 points.

APPENDIXES

A LIST OF POSSIBLE FACTORS OF READABILITY IN BOOKS

295

B "

I. FORMAT or MECHANICAL FEATURES

Size of Book

Average

Larger than a textbook

About 5 by 8

About 14 am* oy 16 cm.

f. Light we

g* Comfortai

light
BTe

h. Not forbidding

I 2. Number of Pages

a. Brief

b.. About 50 pages

o ., Abouic 7JT pages

| 3. Quality cf Paper

Opaque

b Dull surfaced

Evan colored

dTWhite

e Not white

f Pleasant to touch

[ 4. Kind of Type and Printing

a* Large

b. Good-eit eel

e. About 11 pt

About 12-14 pt>

Not under 8

Ola Style

&-
ityle

'4 pte leading^

Spacing, liJfca douole typing apaoe

spaoe

n. Dull ink

o* Clear, legible

296

WHAT MAKES A BOOK READABLE

A. I B

I 5, Length of line

a. About 2 2/S".-2 5/6"

b. Not over 5 1/2"

o 20 pica

6. Margins

D Rather vide

Vide, literal

Medium

Adequate

I 7. General Appearance of Page

a. Not like textbook

o, Broken, not compact

cY Attractive

8. Binding

a. Sturdy, durable

b. Attractive

9* Illustrations

a. Some

Numerous

For biography and science""
"

Adj

biography
cent to t

ext

On same paper as text

Colored

Captioned

Of cartoon type

Wot childish

Appropriate

Attractive

Jffith amps and diagreuns

II. GENERAL FEATURES OF ORGANIZATION

| It Title of Book

a* Striking

[ 2, Chapter Divisions

a. pe$oript;ive chapter neadlngs

b. Running chapter headings

APPENDIXES

297

B C

3. Paragraph Divisions

a* Sub-heads

b. No sub-heads

o. Sub-heads in bold-faced type

d. Interesting sub-heads

a. Not numbered

f . Not like textbook

4. Reference Guides

a. Table or contents

b. Index with catchy title

c No index

d Glossary (not mentioned in text)

e. No marginal notes

f . No footnotes

gf References following text

~h All references in text proper

i Appendix for references

j. Appendix for charts and tables

III. STYLE OP EXPRESSION AMD PRESENTATION

1. Vocabulary

a* Limited

b. Limited to 1000-1500 words

e. Easy

d Not necessarily easy

e. Easy enough for 13-14 year old child

f * Not consciously adapted

g. Simple words

h. Short words

i. Popular

j* Common, familiar

k Non-technical

l. Vivid

m. Anglo-Saxon

n Vernacular, (even colloquial)

^o Dynamic

p Fresh

q. Specific

r. Non-claasical

s* InfonwJ, (noni'-academic)

t Adult (not childish)

2, Sentences

a. Short

b* Varied In length

o. Reasonably short

d* Concise

o. Simple

f Not too involved

298

WHAT MAKES A BOOK READABLE

Without guarded clauses

h. Concrete

l-Bhythoical

Restated in varied ways

3. Paragraphs

a. Snort

D varied in length

c* Reasonable in. length

a. Maximum length of I/? page

e Simple

f. Sue o met

Inviting, arresting

h* Of a sing
i. Prog res si

tnought'-unit

aly continuous

[ 4. Chapters

a. Short

o. Stimulating at beginning
d. Promising at the end

5. Attitude of Author

a. Enthusiastic

D. inspiratlonair

c. Moralizing

d Opt Smi stic

Humanit arl an

f. Baotional, seatimental

6. Method of Presentation

a* Narrative

Biographical

o. Desoriptiv/

d. Not descriptive

Poetic

f Drattatio

g* Joumalistio

7. Style of Presentation

a. Direct

Vivid, colorful

Graphic

d, Lucia, clear

e, Chanaiiag

C. Picture ague

Concrete
1. Apt

APPENDIXES

199

B C

j. Simple

k* Easy, fluent

1. Popular

m* Stimulating

n. Original

o* Distinguished

p, Informal

q. Light, humorous

r Convincing

Stylistic Devices

a. Brevity

b. Omission 1 of non-essentials

o* Simple plot

d. Exaggeration

a. Conversation

f Questions and answers

g. Judicious use of questions and answers

h No questions and answers

i. Familiar verbal illustration

j Some repetition

k* Contrast and comparison

I* Parables

m. Omission of allegory

n. Omission of allusions

o Omission of symbolism

p. Omission of abstractions

q, Accuracy in portrayal

r. No over-specialization

s. Appeal within reader's scale

t Arousal of feeling reaction

u. Adult approach

v. Introduction of new elements singly

vt. Start with familiar

x Portrayal of own personality

y. Realism

e Phantasy

IV-, CONTEM

| I* Then*

&* What people want to read about

b History (especially of IKS+)

o* History (not important)

d f Adventure

0. Scisncc and invention

f. Information

6* People - ^ereonftlities

h, aucoesslui people

i* Idea? of people

j. Not just ideas

fc. Not thecries

1 Romance and action

300

WHAT MAKES A BOOK READABLE

m. Travel and business

. Hiaaan interest

o Seal or ideal life experience

p. Interpretation of life experience

Not analysis of life experience

r . BEtension of numan experienoe

s V Opposed to reality

[a. Mature of Subject Matter

o. unusual

c Familiajr

d. Amusing

e. Homely

f Rich, li-re

Popular

i* Easy

Interesting

k._lfo:

X. He!

Ipflil

n. Purposeful

I g. Unity of Conteat

a.. Single phase""

b. Kot demanding general knowledge

APPENDIX B

Tables LXVI-LXXVII, inclusive, list factors of content, style, format, and
organization ranked of greatest and least importance to readability by each
of three groups of judges. From these tables were obtained the common opin-
ions relative to readability appearing in chapter ii.

TABLE LXVI

FACTORS RELATED TO CONTENT RANKED OF GREATEST AND LEAST
IMPORTANCE TO READABILITY BY LIBRARIANS

Factors Ranked in Highest
Quarter of 47 Factors

Factors Ranked in Lowest
Quarter of 47 Factors

ia Timely subject matter

i| Theme people and personalities

im Theme travel and business

Theme science and invention

Theme romance and action

Theme of human interest

Interesting subject matter

Theme adventure

Content not demanding a general
knowledge

Theme information
ip Theme interpretation of life expe-

rience
ib Theme history (especially of U.S.)

le
i/

*i#
of

*id

i< Theme ideas of people
o.e Homely subject matter
$a Content a single phase
2< Easy subject matter
ik Moral subject matter
ic Theme history not important
ij Theme not just ideas
ik Theme not theories
u Theme opposed to reality
it Theme economics
iq Theme not analysis of life experience
3* Content that adheres to the stated sub-
ject

* Considered of special significance aa indicated on the check-list by ( V \0-

301

302 WHAT MAKES A BOOK READABLE

TABLE LXVII

FACTORS RELATED TO CONTENT RANKED OF GREATEST AND LEAST
IMPORTANCE TO READABILITY BY PUBLISHERS

Factors Ranked in Highest
Quarter of 47 Factors

Factors Ranked in Lowest
Quarter of 47 Factors

id Theme adventure

iff Theme science and invention

in Theme of human interest

la Theme what people want to read

about

ig Theme people and personalities

i/ Theme romance and action

la Timely subject matter

a Humanized subject matter

a/ Interesting subject matter

a/ Helpful subject matter

i/ Theme information
im Theme travel and business

li Theme ideas of people
2k Theme not theories
if Theme self-improvement
2* Homely subject matter
ag Popular subject matter
o.k Moral subject matter
3c Content with clear, prevailing purpose
%d Content in careful sequence
ic Theme history not important
ij Theme not just ideas
iq Theme not analysis of human experi-
ence
is Theme opposed to reality

TABLE LXVIII

FACTORS RELATED TO CONTENT RANKED OF GREATEST AND LEAST IMPORTANCE
TO READABILITY BY OTHERS INTERESTED IN ADULT EDUCATION

Factors Ranked in Highest
Quarter of 47 Factors

Factors Ranked in Lowest
Quarter of 47 Factors

*ia Timely subject matter
im Theme travel and business
I n Theme of human interest

*ih Humanized subject matter
vj Interesting subject matter
3^ Content not demanding general
knowledge

*ig Theme people and personalities
i/ Theme romance and action
la Theme what people want to read
about

*id Theme adventure
I h Theme successful people

*id Purposeful subject matter

ij Theme not just ideas
i k Theme not theories
is Theme opposed to reality
iq Theme not analysis of human ex-
perience

ic Theme history not important
ly Theme imaginative fiction
ix Theme religion and reform
10 Theme health
iw Theme vocational guidance
is Theme understanding the times
lu Theme understanding one's self
3/ Content leading to understanding of
relationships of parts of general sub-
jects

* Considered of special aignificance as indicated on the check-list by (V V)-

APPENDIXES

TABLE LXIX

FACTORS RELATED TO STYLE OF EXPRESSION AND PRESENTATION RANKED OF
GREATEST AND LEAST IMPORTANCE TO READABILITY BY LIBRARIANS

Factors Ranked in Highest
Quarter of 1 16 Factors

Factors Ranked in Lowest
Quarter of 1 1 6 Factors

if Adult vocabulary

8& Use of contrast and comparison

*yd Lucid, clear presentation

jr Convincing style

*i/ Sentences not too involved

8w Start with familiar

8j Appeal within reader's scale

3*r Paragraphs of reasonable length

*5< Enthusiastic attitude of author

Su Adult approach

*ik Non-technical vocabulary

*4^ Chapters stimulating at beginning

2, Sentences varied in length

ya Direct presentation

8y Accuracy in portrayal

1 2 Vivid vocabulary

ij- Informal vocabulary

%h Paragraphs of a single thought-unit

4*/ Chapters promising at end

6a Narrative style

6c Descriptive style

jg Entertaining style

17 Common, familiar vocabulary

i Specific vocabulary

jm Stimulating style

8^ Conversation

2c Reasonably short sentences

3^ Paragraphs varied in length

4^ Clear-cut chapters

la Limited vocabulary

i h Short words

ir Non-classical vocabulary

70 Distinguished style

5/ Emotional, sentimental attitude of
author

8<r Simple plot

I m Anglo-Saxon

i Vernacular (even colloquial) vocabu-
lary

6e Poetic style

7^ Light, humorous style

8jy Realism

3* Simple paragraphs

5jf Truthful, sincere attitude of author

7<r Charming style

jf Picturesque style

8/ Parables

3/ Succinct paragraphs

3<z Short paragraphs

ib Vocabulary limited to 10001,500
words

ie Vocabulary easy enough for twelve-
fourteen-year-old child

4* Chapters of about 30 pages

5<r Moralizing attitude of author

71 Appropriate style of expression
6^ Appropriate presentation

5/ Knowledge of subject

$h Well-balanced attitude of author

%d Exaggeration

8/ 1 Questions and answers

8z Phantasy

* Considered of tpecial significance as indicated on the check-list by ( VV).

3<H

WHAT MAKES A BOOK READABLE
TABLE LXX

FACTORS RELATED TO STYLE OF EXPRESSION AND PRESENTATION RANKED OF
GREATEST AND LEAST IMPORTANCE TO READABILITY BY PUBLISHERS

Factors Ranked in Highest
Quarter of 116 Factors

Factors Ranked in Lowest
Quarter of 116 Factors

3 h Paragraphs of a single thought-unit

is Informal vocabulary

it Adult vocabulary

7# Direct presentation

%q Accuracy in portrayal

ik Non- technical vocabulary

i/ Vivid vocabulary

ib Sentences varied in length

if Sentences not too involved

4-c Chapter stimulating at the beginning

$a Enthusiastic attitude of author

8 Omission of non- essentials

8^ Conversation

8# Adult approach

See; Start with familiar

ic Reasonably short sentences

id Concise sentences

3^ Paragraphs of reasonable length

4^ Clear-cut chapters

j^d Chapters promising at end

6a Narrative style

jd Lucid, clear presentation

8.? Appeal within reader's scale

if Vocabulary not consciously adapted

ih Concrete sentences

3^ Paragraphs varied in length

44 Short chapters

5<? Humanitarian attitude of author

6c Descriptive style

id Vocabulary not necessarily easy

le Vocabulary easy enough for twelve-
fourte en-year-old child

im Anglo-Saxon

ib Vocabulary limited to 1,000-1,500
words

ih Short words

ia Short sentences

$c Moralizing attitude of author

5/ Emotional, sentimental attitude of au-
thor

6d Not descriptive style

6e Poetic style

7<? Charming style

7/ Picturesque style

7*" Apt style

7/ Popular style

jo Distinguished style

Sd Exaggeration

8/ Parables

Br No overspeciaKzation

8* Portrayal of author's personality

8z Phantasy

in Vernacular (even colloquial) vocabu-
lary

ir Non-classical vocabulary

itt Non-technical words explained

ig Sentences without guarded clauses

li Rhythmical sentences

a/ Sentences of varied rhythms

3^ Simple paragraphs

$j Objective attitude of author

7^ Popular style

APPENDIXES

TABLE LXXI

FACTORS RELATED TO STYLE OF EXPRESSION AND PRESENTATION RANKED OF

GREATEST AND LEAST IMPORTANCE TO READABILITY BY OTHERS

INTERESTED IN ADULT EDUCATION

Factors Ranked in Highest
Quarter of 116 Factors

Factors Ranked in Lowest
Quarter of 1 16 Factors

6a Narrative style

ja Direct presentation

ik Non-technical vocabulary

*4C Chapters stimulating at beginning

6c Descriptive style

8 Contrast and comparison

jr Convincing style

Bu Adult approach

*yd Lucid, clear representation

8 Omission of non-essentials

is Informal vocabulary

*i/ Adult vocabulary

4<z Short chapters

qd Chapters promising at end

*5 Enthusiastic attitude of author

jc Graphic presentation

if Vocabulary not consciously adapted

ij Common, familiar vocabulary

2^ Sentences varied in length

*zf Sentences not too involved

ih Concrete sentences

jb Vivid style

Sw Start with familiar

a* Simple sentences

4<J Clear-cut chapters

Sg Judicious use of questions

3^ Paragraphs varied in length

3* Paragraphs progressively continuous

$b Inspirational attitude of author

iff Vocabulary easy enough for twelve
fourteen-year-old child

i h Short words

ir Non-classical vocabulary

2<z Short sentences

la Limited vocabulary

ij Popular vocabulary

7* Apt style

8# Portrayal of author's personality

I b Vocabulary limited to 1,000-1,500
words

id Vocabulary not necessarily easy

3^ Paragraph length about one-half page

3/ Succinct paragraphs

7/ Picturesque style

7/ Popular style

SW Exaggeration

8/ Parables

in Vernacular (even colloquial) vocabu-
lary

a/ Rhythmical sentences

6f Poetic style

8A No questions and answers

5r Moralizing attitude of author

Je Charming style

Jo Distinguished style

5/ Emotional, sentimental attitude of
author

5* Straightforward attitude of author

5Jfe Author's knowledge of reader

5/ Author's knowledge of subject

7/ Natural style

82 Phantasy

* Considered of special significance as indicated on the check-list by ( W )>

306 WHAT MAKES A BOOK READABLE

TABLE LXXII

FACTORS RELATED TO FORMAT RANKED OF GREATEST AND LEAST
IMPORTANCE TO READABILITY BY LIBRARIANS

Factors Ranked in Highest
Quarter of 90 Factors

Factors Ranked in Lowest
Quarter of 90 Factors

8 Attractive binding

q/ Book of about 1 50 pages

9J Appropriate illustrations

4* Type about 1 1 point

4k Well-spaced type

20 Brief book-

9/ Maps and diagrams

4/ Type about 12-14 point

*40 Clear, legible type

4/ Type spaced like double typing

7^ Broken page

5<r Length of line, 20 picas

7* Attractive page

8<: Flexible binding

ib Book of average size

9<? Illustrations on same paper as text

if Light-weight book

9^ Illustrations of cartoon type

3^ Dull-surfaced paper

ic Si2e of book larger than a textbook

402 Black ink

477 Dull ink

9</ Illustrations adjacent to text

9# Illustrations of artistic value

3 Good quality of paper

la Small book

9 Captioned illustrations

ii Book about ao cm. by 14 cm.

9& Attractive illustrations

ih Book of 300-400 pages

i Book of comfortable size

4# Large type

4p Attractive type

dfd Small type

8# Sturdy binding

4r Granjon type

3* Paper not white

4S Roman, Old Style

%c Even-colored paper

50 Length of line about af "-2|"

6d Adequate margins

6/ Well-proportioned margins

4 Good type

9777 Illustrations for travel books

$b Length of line, not over 5 V

6a Wide, liberal margins

* Considered of special significance as indicated on the check-list by (V V)-

APPENDIXES

TABLE LXXIII

FACTORS RELATED TO FORMAT RANKED OF GREATEST AND LEAST
IMPORTANCE TO READABILITY BY PUBLISHERS

Factors Ranked in Highest
Quarter of 90 Factors

Factors Ranked in Lowest
Quarter of 90 Factors

i a
47/2
8
I b
40
ig
4<?
So.

3<:
4/

I/
i h

5<r
6d

9JT
9/

Attractive page

Opaque paper

Black ink

Attractive binding

Book of average size

Clear, legible type

Book of comfortable size

Type about 1 1 point

Sturdy binding

Dull-surfaced paper

Well-spaced type

Appropriate illustrations

Even-colored paper

Open-face type

Attractive type

Broken page

Light-weight book

Book not forbidding in appearance

White paper

Length of line, 20 picas

Adequate margins

Captioned illustrations

Maps and diagrams

ic Size of book larger than a textbook

id Book of about 64-96 pages

4 Type No. 7 Old Style

4 Dull ink

4y Caslon Old Style monotype

4*0 Type harmonious with book

4* Type depending on line and page

$d Length of line about 3 "-4"

5/ Line not over 24 picas

5<? Line about 22 picas

6f Well-proportioned margins

6g- Margins giving balanced effect of

facing pages
\h Glossy paper

le Size of book about 14 cm. by 16 cm.
Q.B Book of about 50 pages
zc Book of about 75 pages
40 Large type
4fd Small type

4J Spacing like double typing
$a Length of line about 2f "~2-{j-"
9* Illustrations in books on biography

and science

o/ Colored illustrations
9& Illustrations of cartoon type

308

WHAT MAKES A BOOK READABLE

TABLE LXXIV

FACTORS RELATED TO FORMAT RANKED OF GREATEST AND LEAST IMPORTANCE
TO READABILITY BY OTHERS INTERESTED IN ADULT EDUCATION

Factors Ranked in Highest
Quarter of 90 Factors

Factors Ranked in Lowest
Quarter of 90 Factors

83 Attractive binding

4772 Black ink

jc Attractive page

ib Book of average size
*4& Well-spaced type
*40 Clear, legible type

yd Illustrations adj acent to text

8<z Sturdy binding

if Light-weight book

9 Captioned illustrations

oj Appropriate illustrations

yk Attractive illustrations

9/ Maps and diagrams

id Size of book about 5" by 8"

3^ Dull-surfaced paper

4p Attractive type

5^ Length of line not over 5 J"
*iA Size of book not forbidding

34 Opaque paper

$d White paper

3j Good paper

9<r Illustrations in books on biography
and science

91 Not childish illustrations

6a Wide, liberal margins

9<? Illustrations on same paper as text

9/ Colored illustrations

ib Book of about 50 pages

ie Book of about 300 pages

40 Large type

3* Not white paper

4*' Type with 4-point leading

5<r Length of line, 20 picas

8^ Appropriate binding

3^ Glossy paper

4^ Small type

4 Dull ink

4/ Type 91 2, point

4# Type with 2-point leading

$d Length of line about 3 "-4"

6e Good margins

8/ Binding in warm, dark colors

ic Size of book larger than a textbook

if Size of book about 14 cm. by 16 cm.

4 A Type No. 7 Old Style

5 Length of line, reasonable

4* Medium-sized type

* Considered of special significance as indicated on the check-list by ( VV)

APPENDIXES

TABLE LXXV

FACTORS RELATED TO GENERAL FEATURES OF ORGANIZATION RANKED OF
GREATEST AND LEAST IMPORTANCE TO READABILITY BY LIBRARIANS

Factors Ranked in Highest
Quarter of 37 Factors

Factors Ranked in Lowest
Quarter of 37 Factors

20 Descriptive chapter headings

4/z Table of contents

la Striking title of book

3f Paragraph divisions not like a textbook

4^ Index with catchy title

4* No marginal notes

4g- References following text

3<? Paragraph divisions not numbered

4/ No footnotes

4* Appendix for references
3* Subheads in bold-faced type
id Dignified title
le Connotative tide
4h All references in text proper
4/ Index
3# Subheads
4? Footnotes

40 Summary of chapters in table
contents

TABLE LXXVI

FACTORS RELATED TO GENERAL FEATURES or ORGANIZATION RANKED OF
GREATEST AND LEAST IMPORTANCE TO READABILITY BY PUBLISHERS

Factors Ranked in Highest
Quarter of 37 Factors

Factors Ranked in Lowest
Quarter of 37 Factors

ia Descriptive chapter headings
4# Table of contents
la Striking title of book
4& Index with catchy title
ic Meaningful title of book
2^ Running chapter headings
$f Paragraph divisions not like a text-
book

4^ All references in text proper
4*' Appendix for all references

Subheads in bold-faced type

No index

Footnotes

Title of human interest

Decorative chapter heads

Progressive chapter divisions

No subheads

Progressive paragraph divisions

Index

310

WHAT MAKES A BOOK READABLE

TABLE LXXVII

FACTORS RELATED TO GENERAL FEATURES OF ORGANIZATION RANKED OF

GREATEST AND LEAST IMPORTANCE TO READABILITY BY OTHERS

INTERESTED IN ADULT EDUCATION

Factors Ranked in Highest
Quarter of 37 Factors

Factors Ranked in Lowest
Quarter of 37 Factors

4<? Table of contents
la Striking title of books
ia Descriptive chapter headings
30 Subheads

3f Paragraph divisions not like a text-
book

4h All references in text proper
%d Interesting subheads
3<r Subheads in bold-faced type
4^ Index with catchy title

ic Meaningful title of book

if Title of human interest

id Headings of chapters to arouse curiosity

if Chapters with concrete introduction

aj- Chapters with summary at end

3^ Paragraph divisions numbered

4k Glossary, mentioned in text

4/ Index

4 Few references

APPENDIX C

The construction of adult reading tests from representative materials of fic-
tion and of general non-fiction is described in Appendix C.

That the tests constructed from the selected materials should represent as
many potential elements of difficulty as possible seemed essential in order,
first, that the criterion of difficulty obtained for each item might truly repre-
sent a combination of many influences; and, second, that the final identifica-
tion of elements of difficulty might be as complete and reliable as the scope of
the study would allow.

Six major steps of procedure were followed:

1. Determining a means of measuring comprehension for this investiga-
tion.

2. Selecting appropriate test items.

3. Preparing responses for the test items.

4. Evaluating the test responses.

5. Rating the difficulty of test responses.

6. Arranging the two tests in final form.

7. Determining the validity and reliability of the tests.

DETERMINING A MEANS OF MEASURING COMPREHENSION

What constitutes comprehension? The type of tests constructed for this
study was determined in the light of the theory accepted regarding the nature
of comprehension. That comprehension is not a single unitary process is gen-
erally conceded. It is, rather, a blending of many processes whose totality
represents understanding. The resolution of comprehension into its compo-
nent processes has not been satisfactorily accomplished. As a consequence,
now one process and now another is taken as a measure of comprehension, as
test-makers try to determine what to measure.

Not only is comprehension thought of as a combination of several processes,
but it is conceived of possessing different levels of quality. From mere percep-
tion rises the beginning of understanding, manifested by thoughts, feelings, or
impulses aroused by the reading material. From this realm of immediate ac-
tion or naive emotional expression, a person arrives ultimately at a level of
understanding which implies some degree of intellectual discrimination, which
requires him to distinguish the thought invited by the words from other
thoughts more or less like it.* It is this level that marks maturity of under-
standing.

* C. K. Ogden and I. A. Richards, The Meaning of Meaning (New York: Harcourt,
Brace & Co., 1927), pp. 185-208*

J. A. Richards, Practical Criticism (New York: Harcourt, Brace & Co., 192,9), pp.
179-88, 326-30.

WHAT MAKES A BOOK READABLE

A thorough measure of comprehension would involve the use of a series of
tests for measuring as many aspects of comprehension as are known. It would
involve, also, a measure of these aspects at various levels, in order that the de-
gree of maturity represented by a reader's comprehension might be deter-
mined. Scientific techniques are as yet too undeveloped for such an ambitious
undertaking. What the test-maker must do, therefore, is to decide upon the
aspect of comprehension he will measure and the level at which he proposes to
measure it. He then faces the baffling question of how to measure it. The is-
sues discussed in the remaining paragraphs of this section indicate something
of the complexity that attends the problem of measuring comprehension.

What aspects of comprehension shall be measured? Two aspects of compre-
hension directed the construction of tests used in this study: (i) the ability to
get the "sense** of what is read, in the form of a general impression of the total
meaning; and (2) the ability to recall specific elements in a selection. These
aspects were accepted as being of first importance, after an examination was
made of the major outcomes resulting from reading general materials.

Since no scientifically determined list of outcomes has been derived, it was
necessary to resort to empirical judgment and to infer probable outcomes from
facts about the purposes for which people read, 2 from evidence as to the mo-
tives that stimulate reading,^ and from typical situations known to promote
various types of reading. 4 An attempt was made to select facts pertaining to
the purposes, motives, and situations that lead to the general reading of books,
magazines, and newspapers. Since the facts themselves are presumably not
exhaustive, no claim can be made for the inferences drawn from them beyond
the probability that they represent the major outcomes resulting from reading
for a variety of purposes in a variety of general situations.

After the facts pertaining to purposes, motives, and situations had been
gathered, the following question was propounded: What outcomes proba-
bly arise from reading "for fun," "for emotional satisfaction," "to ac-
quire general information," "to gather specific information about a special

* Personal conferences with more than 900 adults revealed five major purposes in
reading, as reported in: William S. Gray, 'The Importance of Intelligent Silent Read-
ing," Elementary School Journal, XXIV (January, 1924), 349-52.

From interviews with approximately 300 adults, Parsons obtained a list of frequently
mentioned purposes in reading: Rhey B. Parsons, "A Study of Adult Reading." Un-
published Master's thesis, Dept. of Education, University of Chicago, 1923. Pp. 124.

Interviews with loo adults and 410 answered questionnaires gave Montgomery
data pertaining to the chief reasons for reading recreational material: Wilda Lee Mont-
gomery, "The Investigation of the Uses of Recreatory Reading," University of Pitts-
burgh School of Education Journal, IV (March-April, 1929), 90-91.

5 Interests and motives of approximately 1,200 young people and adults who con-
sulted the reader's adviser in Milwaukee are listed in the following reference: W. S-
Gray and Ruth Monroe, The Reading Interests andHMts ofddufa (New York: Mao-
millan Co., 1929), pp, 268-69.

4 Situations prompting informational and recreational reading appear in: The Twen-
ty-fourth Yearbook of the National Society for the Study of Education, Part I, pp. 5-8.
Bloomington, Illinois: Public School Publishing Co., 1925.

APPENDIXES 313

topic," "to extend experience," "to improve one's literary appreciation," "to
satisfy one's curiosity," or "to forget"?

Four major outcomes were inferred. They may be stated as follows :*

1. The ability to get the sense of what is read in the form of a single im-
pression, evidenced by the statement of a generalization or a summary.

2. The ability to recall specific elements in material, evidenced by a knowl-
edge of the facts presented in it or by the ability to follow directions contained
therein.

3. The ability to apprehend the author's intention, indicated by the read-
er's conception of the purpose that prompts plot and action.

4. An emotional reaction, normally expressed by some personal coloring or
feeling-tone that seems inextricably blended with the sense of what is read.

A critical examination of these outcomes gave rise to two assumptions. The
first was that the ability to grasp the essential meaning of a selection in the
form of a single impression is the outcome that is most frequently demanded
of adults in the reading of general material. The second assumption was that a
large amount of reading is done in order to gain specific information contained
in the selection. Furthermore, since the ability to grasp the essential meaning
seems to depend in a measure upon the ability to react satisfactorily to spe-
cific elements contained m the selection, it was assumed that any technique
devised to measure the first outcome will indirectly measure the second.

Several important questions arose in the consideration of what seem to be
the major outcomes of adult reading. Is the same measure of comprehension
valid for both fiction and non-fiction? If the purposes and motives which
prompt reading "to forget" are different from .those which prompt reading "to
learn," may not the major outcomes of the first be different from the major
outcomes of the second? May not the outcomes of the former tend toward
emotional rather than intellectual reactions, as in the case of the latter?

The implications of such questions as these were weighed carefully at the
outset. It seemed clear from a survey of reading situations that different out-
comes may predominate at different times. Sometimes all four outcomes ap-
pear to function about equally well. Again, all four may fail together, or a
low quality in any one be accompanied by aberrations in the others. If such
observations are well founded, then it seems probable that a direct measure of
one outcome may be an indirect measure of another. Furthermore, the ability
to react pleasurably or unpleasurably to a reading situation implies the ability
to sense in general what is read. No attempt was made, therefore, to meas-
ure emotional reaction, except in so far as that outcome may be positively cor-
related with ability to comprehend the sense of what is read in the form of
a general impression. The technique devised for measuring the ability to grasp
the sense of what is read will be described in the following section.

* Outcomes numbered I, a, and 4 are similar to the major outcomes assumed by
Dale and Tayler as measures of comprehension of technical material, represented by
articles in the field of health: Edgar Dale and Ralph W. Tyler, "A Study of the Factors
Influencing the Difficulty of Reading Materials for Adults of Limited Reading Ability,"
Library Quarterly, IV (July, 1934), 384-412.

314 WHAT MAKES A BOOK READABLE

How to measure comprehension. In this study, the ability to grasp the es-
sential meaning of a selection in the form of a single impression or generaliza-
tion was measured by the success with which an individual identified the gen-
eralization in a series of statements relating to the selection. Two other meth-
ods of measurement, both more direct, were believed unsuited to a testing
situation of the sort created here. The one method, to require the reader to
state orally the impression gained from reading, seemed hopelessly tedious.
The other, to exact a written statement of the general impression gained from
reading, promised to be an invalid measure of comprehension for adults whose
facility in written expression as well as facility in reading may be limited.
Some evidence has been reported by Tyler to show that ability to formulate a
generalization correlates closely with ability to check the best generalization
in a multiple-choice test, as indicated by coefficients of .79 and .85. 6 Accord-
ingly, the latter method of identifying the best generalization, in a series of
statements, was adopted in this study as a valid and expeditious means of
measuring a reader's ability to get the general sense of a selection in the
form of a single impression.

In order to increase the reliability of the measure of comprehension, a sec-
ond reaction to the material was required. This measure, like the first, was
suggested by the Ohio study. 7 The assumption was made that a good test of
comprehension is the success with which one formulates, or identifies, a gen-
eralization of a selection, and that it, furthermore, involves the ability to tell
what is not in the selection or to identify in a series of statements the one that
is false. In other words, if a reader has an accurate notion of what he has read
in a selection, he should also be aware of what he has not read in the selection.
Two reactions were therefore required for each test item as a measure of the
reader's comprehension, defined in terms of the outcomes previously de-
scribed.

Form of test adopted. The multiple-choice test was adopted to measure
comprehension as it has been defined here. This test used a number of discon-
nected paragraphs from reading materials previously selected as sources of
test items. The method of selecting the paragraphs will be explained in the
next section. Two reactions were required for each item to measure ability to
get the general sense of the paragraph and to measure at the same time
ability to recall specific information. The first reaction was the identification
of the best summary of the paragraph in a series of five statements relating to
the paragraph. The second was the recognition, in the series of statements, of
the one not in the paragraph.

SELECTING TEST ITEMS

Test items of fiction and general informational material were chosen from a
large number of sample paragraphs selected from representative books, maga-
zines, and newspapers. The question of what kind of paragraphs to choose

6 Ralph W. Tyler, "Ability to Use Scientific Method," Educational Research Bulle-
tin, XI (January 6, 1932), 1-9.

i Dale and Tyler, op. cit., pp. 6-7.

APPENDIXES 315

from representative reading material to serve as test items was answered by a
consideration of several conditions previously established. These conditions
were concerned with: (i) the outcomes of reading accepted here as measures of
comprehension; (2) the nature of the tests to be constructed; (3) the kind of
adult readers to be tested; and (4) the specific elements whose difficulty was
to be established. Each of these conditions has been discussed in connection
with earlier problems.

From a consideration of these four conditions, four criteria were established
for selecting paragraphs, on the assumption that a paragraph which met all
criteria would meet the conditions of the study and therefore be a valid test
item. The criteria were stated in the form of the following principles:

1. Each item must be independent of the content preceding and following
it in order that responses shall not require further acquaintance with the sub-
ject matter from which the item is drawn.

2. Each item must manifest a completeness, indicated by the development
of a single unit of thought, in order that the reader can get a single impression
of the general meaning of the item.

3. Each item must contain as many variants in expression as possible, in
order that the identification of elements of difficulty in adult reading materi-
als may be reliable.

4. Each item must be brief, in order not to discourage readers of limited
ability, and in order to include a wide sampling of materials. An approximate
length of one hundred words is arbitrarily established.

The selection of "'promising" samples, The first step in choosing sample
paragraphs was to read hastily through a particular book, magazine, or news-
paper, selected as a source of test items, with all of the foregoing principles in
mind. Frequently a cursory glance sufficed to indicate whether a page con-
tained usable paragraphs. All passages that seemed to meet the require-
ments were marked at the first reading. They were then examined critically
with respect to the third principle, which directed attention toward variants
in expression that might influence difficulty. Although it seemed reasonable
to expect that a sampling which claimed to be representative of adult reading
materials would, of necessity, be representative of difficulties inherent in
them, it was believed necessary to make a definite attempt to select items
with a wide range of expressional variants. As a result of such attention in
selecting sample paragraphs, those finally used as test items contained forty-
four of the variants listed in chapter iv in sufficient number to merit correla-
tion.

Passages that appeared to be undesirable after careful scrutiny were
eliminated, until there remained forty-six paragraphs of general informational
content and forty-two paragraphs of fiction from which to choose test items.
The desirability of a passage rested primarily on our personal judgment since
length of paragraph was the only objective standard applied in the selection
of sample paragraphs. Other objective standards were added, however, in
selecting test items from the sample paragraphs.

Analyzing sample paragraphs for elements of potential difficulty. The gener-
al plan of the study called for two tests, one of which would measure compre-

316 WHAT MAKES A BOOK READABLE

hension of fiction; and the other, of general non-fiction. In order that the tests
might be adapted to groups of adults of various levels of ability, the plan fur-
ther required that the tests should represent, so far as possible, a wide range of
probable difficulty.

The next step, therefore, was to arrange the sample paragraphs into ranks
one for fiction and the other for non-fiction, each representing a scale of
probable difficulty, as determined by the occurrence in the items often poten-
tial elements of difficulty. These were chosen somewhat arbitrarily from the
list compiled in chapter iv. There seemed reason to believe, from the evidence
presented in that chapter, that these elements would be the most reliable for
selecting and ranking test items. From the list that follows, it may be noted
that some of the elements had been used earlier in roughly designating the
rank of books, magazines, and newspapers which were to serve as sources of
test items.

The ten elements used in selecting items for the fiction and non-fiction tests
were: percentage of easy words; number of different hard words; number of
words not known to 90 per cent of sixth-grade pupils; percentage of monosyl-
lables; percentage of polysyllables; number of prepositional and infinitive
phrases; number of first-, second-, and third-person pronouns; percentage of
different words; average sentence-length in syllables; and number of e words.

Each of the forty-six sample paragraphs of non-fiction and forty-two para-
graphs of fiction was analyzed for the presence of these ten elements. The
data were tabulated on large tabulation sheets in such a way that each para-
graph, designated by letter, might be seen in relation to every other paragraph
with respect to all ten elements.

Comparison of findings. Each tabulation sheet was examined critically for
the purpose of ascertaining the consistency with which a given paragraph held
to a particular level of relative difficulty among other paragraphs represented
on the sheet. Levels of difficulty were here defined in terms of deviations in
the number or per cent of occurrence of an element in the direction of simplicity
or complexity. If a paragraph showed marked deviation in one or more ele-
ments from its general level, as indicated by other elements, it was rejected.
If several paragraphs held approximately the same place, some of them were
eliminated.

The two tabulation sheets were next examined comparatively for the pur-
pose of determining relative levels of difficulty represented by the two sets of
paragraphs. This examination revealed two important findings: (i) that the
non-fiction material, taken as representative of what adults are reading, ex-
tended farther in the direction of complexity than fiction; and (2) that selec-
tions of fiction reached a level of simplicity lower than general non-fiction. In
order to make the scale of difficulty relatively comparable in the two tests, it
was necessary to improve the "representativeness" of sources of test items by
finding easier non-fiction material of a general informational type and more
difficult fiction.

Extending the range of probable difficulty represented by the tests, Very sim-
ple statements on informational subjects were found in a series of reading

APPENDIXES 317

texts designed for use in adult moonlight schools. 8 These texts are character-
ized by short, frequently repeated words and by brief, simple sentences. Nar-
rative material, of a more complex sort than appeared to be represented by the
samples already analyzed, was found in three novels, The Return of the Native*
Ethan Frame, 10 and Youth. From these books sample paragraphs were ob-
tained and entered on the tabulation sheets at their appropriate levels.

Selection and ranking of test items. Since it was hoped that one value of the
study would lie in its serviceableness in adapting reading materials to abilities
of adults who are limited in their skill in reading, it seemed desirable to utilize
an optimum of test items that could probably be read with some understand-
ing by such persons. A few items whose difficulty was apparently beyond their
comprehension were desired in order that the tests might be used to measure
comprehension of heterogeneous groups. From a critical examination of the
tabulated paragraphs it was possible to select as test items those paragraphs
which seemed to show progressive stages of difficulty over a wide range.
Twenty-five paragraphs in each type of material were selected as test items.
This number was reduced later to twenty-four.

It must be repeated that the arrangement of items in each form represented
merely a scale of relative difficulty, as determined by the occurrence often po-
tential indicators of difficulty. Such an arrangement may seem to disregard the
influence of other elements which had been set off as bearing probable relation
to difficulty. Since the entire list of elements compiled in chapter iv had been
kept in mind in choosing sample paragraphs, it was believed that the definite
implications of a few "promising" elements, whose relationship to difficulty
was already known for other types of material, were adequate for the final se-
lection of test items.

PREPARING RESPONSES FOR TEST ITEMS

After selecting test items, the next important step was the preparation of a
series of responses for each item from which the reader was to choose the one
he thought the best summary of the selection and the one he thought the
poorest.

Number of responses. In every case five responses were formulated, as
shown in the sample item on page 60. These represented the best summary
statement of the thought in the paragraph; the worst summary statement,
presenting something not in the paragraph; and three others, in approximately
normal distribution between the two extremes. The use of at least four or five
responses was believed to be distinctly advantageous in minimizing chance

8 Cora Wilson Stewart, Country Life Readers ', Books I and II. Richmond: Johnson
Publishing Co., 1931.

* Thomas Hardy, The Return of the Native (New York: Charles Scribner's Sons,
1917), p. 116.

M Edith Wharton, Ethan Frome (New York: Charles Scribner's Sons, 1911), p. 142,
" Joseph Conrad, Youth (Garden City: Garden City Publishing Co., 1903), p. 192.

3i8 WHAT MAKES A BOOK READABLE

successes in multiple-choice tests. 13 Another advantage was believed to attend
the use of five responses in the present study. It seemed probable that some
adults could draw conclusions closer to either or both of the correct answers
than could others. The finer measuring device would therefore provide a means
of assigning credit for differences in ability with relative fairness and dis-
crimination, the credit being determined by the reader's ability to choose the
best and the poorest summary for each item.

Formulating summary statements. The method followed in obtaining re-
sponses to the test items was to have three competent readers formulate two
tentative summaries for each paragraph in the two tests. Each reader was
asked to do three things:

I. To read each test item and then to formulate independently what he be-
lieved to be the best and the poorest summary of the thought in the para-
graph. Caution was given to make the poorest summary sound plausible so
that its poor quality would not be obvious.

a. To state each summary in language less difficult than the test item.
This precaution was taken in order to prevent the later assignment of a degree
of difficulty to an item that was truly the product of the difficulty for its re-
sponses.

3. To evaluate the summaries in joint conference with the other two read-
ers and to select the best statements of the best and poorest summaries to use
as test responses.

The three intervening responses between the best and the worst sum-
maries were formulated later with the help of two of the readers. We then
determined the order of arrangement of the five responses for each item by
chance.

Preliminary evaluation. Before the responses were accepted as final, the
tests were given to adult groups to evaluate the responses. Two purposes
prompted this step in the procedure. They were: to determine how well the
responses had been scaled from best to poorest, and to discover the extent to
which the wording of the responses involved difficulty.

The tests were given in mimeographed form to an undergraduate class at
the University of Chicago, whose members were majoring in English and the
Social Sciences, and to a graduate class in Education, meeting at the Univer-
sity College for a course called "The Use of Achievement Tests for Improve-
ment of Instruction/' The following instructions were attached to the test-
sheets:

"Read over the twenty-five items in each test. After you have read an
item, assign values to each of its five responses. Give a value of / to the re-
sponse that is the best summary of the paragraph. Give 5 to the poorest sum-
mary. Distribute scores for the other three responses as you think they should
be evaluated. If you believe a response differs only slightly from the best, in-
dicate by a score near i, possibly 1.5. If you believe two responses are equally
good summaries, you may give each of them a ranking of i.

" Similarly, if you believe one response is very close to the one you rate the
lowest, you may give it a score of 4.5, or 4.8, or whatever your judgment die-

M G, M. Ruch, The Objective or New-Type Examination (Chicago: Scott, Foresman
&Co.,

APPENDIXES 319

tates. Also, if you think two responses are in all ways equally unsatisfactory
you may rate each of them 5.

"After you have assigned values to each of the five responses, will you in-
dicate how much difficulty you had in deciding which was the best response?
And in deciding which was the poorest response ? Do this by writing E for easy,
M for medium, and H for hard on the lines provided for this record."

The administrator of the test took opportunity while the students carried
on this evaluation to observe the amount of time utilized in reading the par-
graphs and to estimate the amount needed to select the best and poorest re-
sponses. The evidence seemed to indicate that adults whose abilities were
similar to those of university students could finish either test in a period of
twenty or thirty minutes and that adults of lesser ability would probably re-
quire a longer period for answering items within their comprehension.

Revision of test responses. Approximately thirty-five students evaluated all
of the responses for both tests. Their reactions were tabulated and used as a
basis for revision of the test responses. Two types of revision were made, one
intended to improve the scaling of the responses from best to poorest sum-
maries, and the other to obviate any difficulties discovered in determining
which response was the best or the poorest summary.

In revising responses to improve the scaling, the following plan was used.
If considerable disagreement was noted in ranking a given response, the state-
ment was either modified in the direction of its most frequent ranking or it
was rejected entirely and a new statement formulated.

In revision designed to overcome difficulties in choosing the response stat-
ing the best or the poorest summary, an effort was made to obviate the tend-
ency to test judgment above comprehension. For example, if university stu-
dents marked a summary statement H, it was evident that they had difficulty
in selecting which was the best response in the series. If, in addition to indi-
cating the hardness of selecting the response, they also rated a poor summary
as best, or rated more than one response as best, it was evident that the best
response was not discriminative. In such a case, a new best response was for-
mulated which aimed at a higher quality of goodness as a summary statement
than other responses in the series.

EVALUATING TEST RESPONSES

Theory of evaluation. After responses of doubtful value were revised and
critically examined by the readers who had first formulated them, the next
step was to evaluate each response in order to determine the credit it should
receive as a measure of comprehension. That is, it was essential that each re-
sponse in a series should have a value, which would represent the amount of
credit that an individual should receive for marking it the best and the amount
that he should receive in case he marked it poorest. There was believed to be
greater fairness attached to giving credit for every response than would attend
giving credit for only the responses established as statements of the best and
the poorest summaries. As was mentioned in the previous section, an attempt
was made to scale the responses according to the goodness of the summary
represented by them. It was assumed, therefore, that for every item there
was a second response closer to the best than any other; and also a second re-

320 WHAT MAKES A BOOK READABLE

sponse closer to the poorest than any other. There was presumably an inter-
mediate response, as close to the best as to the poorest.

If an individual should choose a second-rate response, he should receive
credit indicating better comprehension than would have been indicated by
choosing the intermediate response as "best" or ^poorest" Although assign-
ing credit for every response (which would account for varying degrees of
merit) was less expeditious than assigning credit for only the best and the
poorest summaries, it seemed a more valid measure of comprehension.

Method of evaluation. In order to determine the worth of each response as
a paragraph summary, nine persons who had had considerable experience with
tests were asked to rate the five responses for each test item. Copies of both
tests with their revised summaries were given to each reader. The same direc-
tions were given to them as had been given to the university groups, namely,
(i) to signify the best response by a rating of i, the poorest response by a rat-
ing of 5, and other responses by a rating which they judged valid; and (2) to
indicate for each item the degree of difficulty experienced in selecting the best
and the poorest responses.

When the ratings assigned to the five responses on each item in the tests
were tabulated, practical unanimity was found among the judges in their rat-
ing of the best and the poorest response. Some disagreement, the extent of
which will be discussed in succeeding paragraphs, was noted in the rating of
other responses.

From the independent ratings of each response by nine judges, a value was
calculated to be used in determining the credit received by any individual
checking that response as best or poorest. The simple average of the nine rat-
ings was taken as the mean value.

An examination of these mean values showed that, in many cases, the mean
value of the best and the poorest response was i and 5, respectively. These
values usually represented the average of identical ratings for these responses
by all judges. Where such was the case, these values were considered "true"
values of the responses in so far as the identical rating of the nine judges may
be taken as representative of the average rating of an infinite number of
judges.

Concerning the variability among ratings of other responses, the first ques-
tion raised was : How well does the mean value of each response represent the
true value which would be obtained from an infinite number of judges? The
method followed in answering this question was to express the reliability of a
mean value in terms of the standard deviation of the individual ratings for the
response, divided by the square root of the number of judges rating the re-

sponse; that is, in terms of CTM=-~~ - 13 The meaning of standard errors of

** Henry E. Garrett, Statistics in Psychology and Education (New York; Longmans,
Green & Co., 1926), p. 121.

Since the formula for 0"(dis) 3 ='c|"-:rr- , the formula for &M becomes

_
bO>

*\f w

APPENDIXES 321

the mean values of the responses, obtained by this formula, may be illustrated
as follows: 14

Take response Number I in Item I, Form I. It was rated 2 by six judges
and 3 by three judges. Its mean value is, therefore, 2.3. It has a standard
error of .156, which means that the probable chances are two out of three that
the true value lies within one standard error of 2.3, that is, within the limits
2.3+. 156 and 2.3 .156, or between 2.456 and 2.144.

The standard errors were used here to determine the significance of differ-
ences between adjacent values. Again, Item i, Form i, may be taken as an
example. A comparison of the mean value for response Number i, which is
2.3, with the mean value for response Number 3, which is 2.7, indicated that
the one is not markedly better than the other. Since the standard errors, rep-
resenting their deviations from the true values, are low, it was evident that the
responses are more sharply evaluated than is shown by their mean values
alone. The amount of difference between adjacent responses in other items
was interpreted in a similar fashion.

Reliability of judgments* Another approach to an evaluation of reliability
was through the question: Are nine judges enough so that their composite
mean values on one item are statistically reliable? To answer this question,
one needs to know the average intercorrelation among the judges on each
item and then apply the Spearman-Brown Prophecy-Formula to estimate the
number of judges needed to attain a minimal reliability desired for a given
item. 15 Frequently only an approximation to the average intercorrelation of
judges is used, either the correlations between the ratings of two of the judges,
randomly chosen, or an average of three or more intercorrelations from the
total number. Rather than make use of such approximations, based on a few
of the judges, we decided to make approximations based on all the judges
for a part of the items. The formula used for solving the average intercorrela-
tions makes use of actual ratings without computing any of the individual
correlation coefficients. 16 Since the use of this simplified formula still in-
volves a tremendous amount of labor for all items in the two tests, it was be-

*4 For individual ratings by the nine judges, the mean values, and the standard errors
of the mean values, see: Bernice E. Leary, "Elements of Reading Materials Contribut-
ing to Difficulties in Comprehension on the Part of Adults" (Doctor's dissertation,
Department of Education, University of Chicago, 1933), pp. 97-104.

s Truman L. Kelley, Statistical Method (New York: Macmillan Co., 1924), p. 218.
16 The formula is stated as follows:

ss~ ; A fj I <_j i I i~"~~^v I *J I I |~ f , i** mraiav*** iT ~~ u*t .uujuiubj 0* TC'*

spouses; number of judged; $ means a summation for each response of : where;

Ay'B^C / judges; - for one judge. See: Harold A, Edgerton and Herbert

A. Toops, "A Formula for Finding the Average Intercorrelation Coefficient of Unranked
Raw Scores without Solving Any of the Individual Intercorrelations,*' Journal of Edu-
cational Psychology, XIX (February, 1928), 131-38.

322

WHAT MAKES A BOOK READABLE

lieved that the average intercorrelation of selected items from each form would
give results which might serve as a basis for estimating the reliability of the
composite judgment for all items.

A range in reliability of judgments was obtained from average intercorrela-
tions computed for items whose variability was greatest and least for all
items, as indicated by their average standard errors. Items numbered I and
15 had the least and the greatest variability, respectively, in composite judg-
ment for all items in Form i, Item Number 6 having been cast out of the
test for reasons that will be explained later. Among items in Form 2, those
numbered, respectively, 19 and 23 had the least and the greatest variability,
excluding Item 15, which had been rejected from the test along with Item 6 in
Form i.

The average intercorrelations were computed for these items and their
values substituted for TV in the Spearman-Brown Prophecy-Formula to ob-

TABLE LXXVIII

RELIABILITY or COMPOSITE JUDGMENT FOR ITEMS REPRESENTING THE
RANGE OF VARIABILITY IN FORM i AND FORM 2

FORM

Num-

Average

Num-

Average

Item

ber of
Re-

Intercor-
rela-

Relia-
bility

Item

ber of
Re-

Intercor-
rela-

Relia-
bility

sponses

tions

sponses

tions

.004.

.OK

I r

,OOQ

.0*76

y* j
.784

070

tain their reliability coefficients. 17 Results are shown in Table LXXVIII. The
following conclusions may be drawn from the facts presented in this table:

1. The reliability of composite judgments of nine judges for both forms of
the test ranges from .994 to .970, when items with lowest and highest variabil-
ity in judgment represent the range.

2. There is evidence that the nine judges used here would agree closely
with another group of nine judges as to the relative value of test items in
Form i and Form 2,

3. The nine judges agreed more closely with another group of nine, as indi-
cated by the reliability coefficients, than their mean values agreed with the

r ? Formula for prediction is

See: Holzinger, op. cit^ p.

APPENDIXES 323

true values of responses (other than best and poorest), as indicated by the
standard errors.

4. For purposes of this study, nine judges were reliable, since perfect reli-
ability of the composite score would, of 'course, be possible only by using an
infinite number of judges.

So far as reliability could be determined by the methods reported in this
section, it was believed that the mean value of each response, representing the
combined ratings of nine judges, might be taken as reliable measures of value.

RATING DIFFICULTY OF RESPONSES

Method used. In addition to evaluating each test response by ratings from
i to 5, the judges were asked to indicate by the letters -E, Af, and H> the degree
of difficulty which they experienced in discriminating the best and the poorest
response of each item. This difficulty is not the same as difficulty in compre-
hending the reading materials, since it is due to the wording of the response
rather than to the meaning of the selection.

A somewhat arbitrary method, designated the 2-1-0 method, was utilized
to give numerical values to the difficulty ratings. Each H rating was given 2
points; each M rating was given I point; and each E rating, o points. All
the ratings of the nine judges were then tabulated by points, for the best and
the poorest response in each paragraph. From these the total difficulty of each
response was computed. For example, the best response for Item i, Form i,
was rated E by eight judges and M by one judge. No judge rated it H. Its
summated difficulty was therefore 8 Xo plus i X I or r. The poorest response
for the same item, rated E by six judges, M by two judges, and H by one
judge, had a difficulty of 4.

Validity of the 2-i~o method. Although the a-i-o method of rating diffi-
culty is clearly arbitrary, it has been found to give results about identical with
those of the deviation method/ 8 which calculates values for an individual's
ratings of E, M> and H, separately, in terms of the standard deviation from
the individual's mean rating.*? Since the a-r-o method of calculating difficul-
ty scores thus appears to be valid, it was used in preference to the deviation
method on account of its simplicity.

Interpretation of difficulty values. From the relative values of E, M, a.ndH
assigned to each response, it was possible to determine whether variations in
difficulty of choosing the best and the poorest responses were of enough sig-
nificance to require weighting the value of the selections. Accordingly, the nu-

* 8 Douglas Waples and Ralph W. Tyler, What People Want to Read About (Chicago:
University of Chicago Press, 1931), pp. 173-74-

** The standard deviation value for an individual rating is calculated by the formula:
Ti/t V V A*f

^ " in which * signifies the mean value for a given rating expressed in
(T r ff

standard deviation units; Yi represents the left ordinate of the portion of the assumed
normal curve which thia rating represents; Y* represents the right ordinate; and F signi-
fies the fractional part of all the individual's ratings which are a, I, or o, as the case
may be. Holzinger, op. cfa 9 p. an.

3 2 4 WHAT MAKES A BOOK READABLE

merical values of E y M, and H were thrown into two frequency distributions
for each test, one for the numerical difficulty assigned to choosing the best re-
sponse; and the other, the poorest response. The modal value was then found
by inspection to serve as a standard. A deviation of four points in either direc-
tion from the mode marked a limit above or below which the difficulty of a re-
sponse was plainly distinct from the difficulty of other responses in the test.
Approximately all of the best and poorest responses in both tests were found
to deviate relatively little from the mode. This fact indicated that difficulty
due to wording of the responses might be considered generally insignificant.
Two exceptions were noted in Item 6 of Form i and in Item 15 of Form 2.
In both cases, the best and poorest responses deviated four and six points
from their respective modes, indicating that these responses contained inher-
ent difficulty due to wording.

The usual method of correcting for difficulty is weighting the response in
the direction determined by the deviation. Although technically the weight-
ing of test responses insures greater accuracy, practically weighted scores have
been found to correlate so closely with unweighted scores that the work in-
volved in deriving and using weights is considered unnecessary. 30 Since the
weight of a series of scores is given by its variability, a series which has not
been artificially weighted may be said to have a natural weight corresponding
to its standard deviation.

The process of weighting having been shown to be not only cumbersome but
more or less futile, it was decided to reject Item 6 from Form i and Item 15
from Form 2. With these omissions, the number of test items in each form
was reduced to twenty-four.

ARRANGING THE TESTS IN FINAL FORM

With the rating of test responses for difficulty, the last major task in the
construction of the tests was completed. A set of directions and a fore-exer-
cise preceded the first item in each test form. They have been illustrated earli-
er in chapter iv.

Provision was made for obtaining personal data about the subject in spaces
designated for that purpose on the cover page of each form. These data in-
cluded name of subject, date of testing, age within nearest five years, last
grade attended in day school, length of attendance in night school, amount of
time devoted to reading per day, and name of occupation.

The arrangement of items in order of their probable difficulty has been de-
scribed in a previous section. The tests were printed in booklet form to in-
sure legibility and ease of handling. 21

* Evidence for this statement is reported in the following references: Karl R. Doug-
lass and Peter L. Spencer, "Is it Necessary to Weight Exercises in Standard Tests?"
Journal of Educational Psychology, XLV (February, 1923), 109-12; C. W. Odell, "Fur-
ther Data Concerning the Effect of Weighting Exercises in New-Type Examinations,"
#/., XXII (December, 1931), 700-704; Douglas E. Scates and Forest R. Noffsinger,
"Factors Which Determine the Effectiveness of Weighting," Journal of Educational
Research, XXIV (November, 1931), 280-85.

* Copies of Form i and Form 2 of the Adult Reading Test appear in Leary, op. cif.

APPENDIXES 325

SCORING TEST ITEMS

The general plan adopted in the present study, as presented earlier, pro-
vided that varying degrees of merit were to be assigned to each item, accord-
ing to the value of the responses indicated by the reader as the best and the
poorest summary. This method of giving some credit for any response checked
by the reader was believed to give a more discriminating measure of compre-
hension than the easier method of crediting only the selection of the truly best
and poorest summaries. In accordance with this plan, values were derived for
each response in the test items. Knowing the value of each response, we were
able to calculate a reader's score on a test item by the method described in the
following paragraph. 23

For any response designated as poorest by "o," a score was given which
represented the value previously obtained by averaging the ratings of nine
judges. From this score was subtracted the value of the response designated
as best by * l< V " Hence the score on any test item was the difference between
the value assigned to the poorest response and the value assigned to the best.
Had all of the nine judges been in perfect agreement as to the rating of the best
and the poorest responses, it may be seen that the best response would have
always a value of I, and the poorest, a value of 5. A reader who checked both
responses correctly would receive a score representing the difference between
5 and i, or 4. The reader who checked either or both responses incorrectly
would receive a score less than 4. Actually, however, the average value as-
signed to any response in an item was frequently not an integer, but a decimal
fraction.

To illustrate how the method of scoring was used, a test item from Form 2
will be taken as an example. In this item the best summary was given a value
of i.i; the second best, 1.6; the third best, 3; the fourth best, 3.6; and the
poorest, 5. If a reader checked both the best and the poorest responses cor-
rectly, his score would be 5 minus i.i, or 3.9, the highest score obtainable on
that item. If he checked the worst response correctly, but checked the second
best as best (a circumstance that could easily occur since the two responses had
been closely evaluated), his score would be 5 minus 1.6, or 3.4. If he checked
the worst response correctly, but chose the third best as best, his score would
be markedly lower, being 5 minus 3, or a. If he made the worst possible choice,
that is, if he checked the poorest response as best and the best as poorest, his
score would be i.i minus J, or 3.9. It may be seen that the application of
this method gives a possible range of scores for any test item from a theoreti-
cal +4 to 4, with, many possible intervening scores expressed either integral-
ly or decimally.

ARE THE TESTS VALID?

Probably the most important single measure of the value of a test is the de-
gree of validity possessed by it. Since the validity of the facts secured depends
upon the validity of the measure used to obtain them, it is obviously true that

** We are indebted to Professor Ralph W. Tyler, of the Bureau of Educational Re-
search, Ohio State University, Columbus, Ohio, for the method of scoring described
here.

326 WHAT MAKES A BOOK READABLE

if a test does not measure what it purports to measure, the facts obtained
through its use are worthless for their purpose. Unfortunately, it is impossi-
ble, or at least extremely difficult, to prove conclusively by direct methods
that a test does or does not measure that which it is intended to measure.

Two difficulties were encountered in determining the validity of the tests.
The first lay in the conditions attending the testing of adults, discussed in
chapter iii. The second lay in obtaining a suitable criterion with which to cor-
relate the results of the tests constructed for use in the study. Since it is sel-
dom possible to carry on an extensive testing program with adults, it was
necessary to use the time allowed for the investigation to the best advantage.
Whatever test was taken as a criterion had of necessity to be brief. Further-
more, the heterogeneity of the groups tested demanded that the test should
have norms for different adult levels of ability. An additional requisite was
that the criterion measure the same aspect of comprehension as was measured
by the tests constructed in this study.

The Monroe Standardized Silent Reading Test, although not altogether
satisfactory, seemed to fit best the conditions of the situation. Since but four
minutes are required to take the test, it could be administered on the same
occasion as the Adult Reading Test. Moreover, the Monroe test is so con-
structed that the various forms of Test I can be used for grades 3, 4, and 5;
those of Test II, for grades 6, 7, and 8; and forms of Test III, for grades 9 to 12.
The scores on all the tests can then be transmuted to a common basal scale,
represented by "B" Scores, which designate the grade level achieved in
reading.

The aspect of comprehension measured by the Monroe test is the ability to
secure information from reading content, 33 The ability is indicated by under-
lining or writing words according to directions. In the Adult Reading Test,
comprehension is measured by the ability to grasp the general sense of
what is read in the form of a summary statement. This ability was assumed to
depend in a degree upon the ability to obtain specific information. A measure
of the one ability, therefore, was believed to give a measure of the other. It is
clear that the aspect of comprehension measured by the Monroe test is not the
same as that measured by the Adult Reading Test. Whatever conclusions are
presented regarding the validity of the latter test are made with certain reser-
vations.

In the first place, a single criterion may give accurate evidence of validity
of the test with which it is correlated only if the criterion itself is valid. No
data are available pertaining to the validity of the Monroe test for testing
reading comprehension of adults, although some evidence has been compiled
relative to its validity at lower levels. Pressey, for example, found that com-
prehension, as measured by the Monroe test, correlated .27 with teachers'
estimates of reading ability by a pooled rating. However, the validity of
teachers* estimates as criteria was unknown. 2 -* Gates secured evidence of

3 * Walter S. Monroe, "Monroe's Standardized Silent Reading Tests," Journal of
Educational Psychology, IX (June, 1918), 303-12.

3 S. L. Pressey and L. W. Pressey, "The Relative Value of Rate and Comprehension
Scores in Monroe's Silent Reading Test, as Measures of Reading Ability," School and
Society, XI (June 19, 1920), 747-49-

APPENDIXES 327

greater validity by correlating the Monroe test with a series of other standard-
ized reading tests. 25 He found a mean relationship of .72 with the Burgess Pic-
ture Supplement Test, of .60 with the Thorndike-McCall Reading Test, and
of .75 with the composite comprehension tests, when correlations were based
on scores for grades 3-6. He concluded that the correlation of .60 with the
Thorndike-McCall test might be accounted for by the fact that each measures
a quite different type of comprehension or measures it in a different way.
Traxler's comparative study of the reading ability of junior high-school stu-
dents indicated a relationship of .66 between the Monroe Standardized Silent
Reading Test and the paragraph-meaning part of the Stanford Achievement
Test; and a correlation of .56 between comprehension on the Monroe test and
the Thorndike-McCall.* 6 Evidence presented from these studies indicates
that comprehension, as measured by the Monroe test, is not closely related
with comprehension, as measured by other standardized tests, for lower levels
of ability.

A second reservation that needs to be made in interpreting the data pre-
sented in this section has already been indicated. Low correlation between
comprehension tests may be caused not by low validity but by the different
aspects of comprehension which they measure. Correlations between the
Monroe test and the Adult Reading Test, therefore, only qualifiedly indicate
validity.

Comprehension scores of 756 adults were secured for the Monroe test and
for both forms of the Adult Reading Test. The scores made on the first were
then correlated with the scores made on Form I and Form 2 of the second.
Table LXXIX presents the coefficients of correlation obtained for twelve
combinations of groups. 27 It may be noted that in this table, evening-school
groups, numbered i, 2, and 3, were taken together, as were parent-teacher
organizations, number 20 and 21, and other groups of a relatively homogeneous
composition. While all but one coefficient of correlation, shown in Table
LXXIX, are positive and significant, being more than four times their prob-
able errors, they are not high enough to indicate close agreement between the
two tests. Although the disparity in relationship may be due in part to the

*s Arthur I, Gates, "An Experimental and Statistical Study of Reading and Reading
Tests," Journal of Educational Psychology, XII (November, 1921), 445-64.

36 Arthur Edwin Traxler, "The Measurement and Improvement of Silent Reading
at the Junior-High-School Level" (Unpublished Doctor's dissertation, Department of
Education, University of Chicago, 1932), p 85.

a * Group numbers refer respectively to: Dante Adult Day School; Englewood Eve-
ning School; J. Sterling Morton Evening School; Glenn Street Evening School; Berea
Foundation Junior High School; Agricultural and Mechanical Arts Model School, grade
7; the A and M Model School, grade 8; the A and M Model Junior High School,
grade 9; the A and M Model Senior High School, grade 10; the A and M Model
Senior High School, grade 1 1; the A and M Model Senior High School, grade 1 2; Berea
Academy; Martha Berry Senior High School, grade 9; Martha Berry Senior High
School, grades 11-12; Rabun-Gap Nacoochee Mountain School, grade 10; Rabun-Gap
Nacoochee Mountain School, grade II; Berea College; Martha Berry College, Fresh-
men; Martha Berry College, Juniors ;Hinsdale Parent- teachers' Association; and Con-
gress Park Parent-teachers' Association.

3*8

WHAT MAKES A BOOK READABLE

unreliability of both tests, it is probable that they test somewhat different
abilities. The fact that working times are very different in the two tests may
also influence their validity. Since the Monroe test was administered before
the Adult Reading Test, and since it is a timed test, it undoubtedly had the
disadvantage. This seems likely in the case of adults who had never before
taken a new-type test and for whose groups the validity was especially low.

Another means used to determine the extent of agreement between the two
tests was in connection with the selection of "poorest" and "best" readers.

TABLE LXXIX

CORRELATIONS BETWEEN SCORES ON FORMS i AND 2 OF THE ADULT READING
TEST AND THE MONROE STANDARDIZED SILENT READING TEST*

FORM i

FORM 2

Group

Number
of Cases

r P.E.

Group

Number
of Cases

r P.E.

4 f .

ic. 16 . .

AJQ

. 64.0 4- 062

jr. 1 6

4.O

. <QI + .060

.C-d.7

2O, 21

.4.08+ .oc6

2O. 21

.4.CQ4- .OCQ

C . . . .

4.86+ OCA

A7A"t OC8

j
Q, IO, II

.4?8 .061

yw
Q7

4.1 Q 4* OC7

I 2. 1

4.27 + 068

A A

orr-h 088

44.

.AI7-I- .084.

18. IQ ,

7CO+ 062

6,7, 8

.T7Q+ .084.

14. T

14.

726

18. IQ

. 7?<-fc 06?

I* 2. 7

7O8+ O7C

I4Y

.772

Q ICX II ...

267+ .074.

*W"

.2OC + OO2

2C2+ 067

.276-!- .067

6, 7, 8

y u

4-8

.182+ .OQC

. eit., pp. 66, 83.
f Probable errors for these groups were not computed because of the small number of cases.

When the "B" Scores obtained on the Monroe test by 756 adults were distribu-
ted, it was found that of the 190 adultswho ranked below ^ x on the Adult Read-
ing Test, 109, or about 58 per cent of the group, ranked below j^ on the Monroe
test. Of the 191 who ranked above ^ 3 on the Adult Reading Test, 98, or about
52 per cent of the group, ranked above ^ 3 on the Monroe test. This fact shows
that '^poorest" and "best" readers on one test tended to be so designated on
the other, but again the agreement is not significant enough to indicate more
than positive relationship between the two tests.

The validity of test items. In the selection of test items for a test intended
to measure a particular ability, one of the primary concerns of the test-maker
is to secure individual items that are valid or discriminating. The discriminat-
ing power of a single test item refers to the degree to which success on that
item by itself indicates possession of the ability which is being measured. An

APPENDIXES 329

item may be said to be perfect in discriminating power, when every individual
who scores successfully on the item ranks higher on a scale of ability than any
individual who fails the item. An item is said to have zero discriminating pow-
er when there is no systematic difference between the ability of the individuals
who succeed on the item and those who fail. Items of all degrees of discrimina-
tion may be found between the extremes of perfect and zero discriminating
power. 38

Many suggestions are given in the literature of test construction relative to
procedures for measuring the "goodness** of a single item. They range all the
way from bi-serial r, which shows the relationship between success or non-
success on the item and the criterion measured, to makeshift devices, which
may approximate an accurate estimate of '"goodness." Although bi-serial r
gives the best index of the discriminating power of test items, since it is based
upon all the data from a group, it has the disadvantage of complexity of calcu-
lation not found in empirical methods.

In the present study the procedure used was adapted from a simple one
used in a study by Traxler at the suggestion of Professor Karl J. Holzinger, 29
Two assumptions were made. The first has already been made in this investi-
gation. If the test papers of a group are divided into three classes, comprising
the highest 25 per cent, the middle 50 per cent, and the lowest 25 per cent of
the total scores, then the students in the upper 25 per cent are relatively su-
perior and those in the lower 25 per cent are relatively inferior readers (provid-
ed that the entire test is valid). In the present study these two classes of read-
ers have been designated, respectively, as ^best*' and "poorest" readers. The
second assumption was that if a single test item is valid, that is, if it has dis-
criminating power, most of the "best" readers will answer it correctly and
most of the "'poorest*' readers will answer it incorrectly. Since every test item
in this investigation was given some value, ranging from +4 to 4, "best* 1
readers should receive higher scores on a discriminating item than "poorest**
readers.

In the following formula, adapted from Traxler's study, the terms and
their interpretations have been fitted to the conditions of this study: 30

8 E. F. Lindquist and Walter W. Cook, "Experimental Procedures in Test Evalua-
tion," Journal of Experimental Education, I (March, 1933), 163-85.

a > Traxler, op. cit.> p. 85.

3 The original formula, based on definitely designated right (R) and wrong (W) re-
sponses of the upper () and lower (/) 25 per cent of the readers, was stated:

~ N

ibid. y p. 88.

WHAT MAKES A BOOK READABLE

in which V is the validity of the test items, $b is the actual score of "best"
readers, D& is the difference between the possible score and the actual score of
"best" readers, S P is the actual score of "poorest" readers, D p is the difference
between the possible score and the actual score of "poorest" readers, and T is
the total possible score of "best" and "poorest" readers.

By this formula the index of discrimination may vary from +1.00 for the
item with highest discriminating power to i.oo for the item with lowest dis-
criminating power. An item which is answered as well by "poorest" read-
ers as by "best" readers has an index of validity of "o."

The V of each of the twenty-four test items in Forms i and 2 of the Adult
Reading Test is shown in Table LXXX. The entries indicate that all of the

TABLE LXXX

VALIDITY OF THE ITEMS IN FORMS i AND 2
OF THE ADULT READING TEST

Item

Formi

Forma

Item

Form i

Form 2

666

T7C

600

642

.4^C

'of j
.617

14.

v ^y

. ceo

* U T

.-108

-J

.7OC

."ta7

c64

. CO7

.440

4OC

C77

67O

.484

CO I

17 ....

647

606

4.28

788

18 .

8T*

86c

6l7

4Q2

610

j
6c8

.564

Vy*
.2Q7

.602

.7ar

.652

.606

.68c

.701

.86?

.4.IO

.74.0

.T77

II ..

.607

.760

.671

.776

12 .

.8oc

co6

24.

712

.6 co

items have considerable validity, ranging for Form i from .867 to .428; and for
Form a from .865 to .297. There was generally a marked difference between
the performance of the "best" readers and the "poorest" readers. Exceptions
may be noted in the case of items near the beginning of the tests, which were
intended to be so easy as to be comprehended by all adult readers. That one
item did not prove this easy is shown by the relatively high V for Item i,
Form i. Since this item was selected because it contained a low percentage of
elements of difficulty, a low ^was anticipated. A probable explanation for the
unexpected discriminating power of the item is the fact that the choice of the
best response required, in addition to reading comprehension, the ability to
calculate in mathematical terms from the content of the item that "I spent
about five months making a boat I needed out of a big tree*" An analysis, of
responses to this item seemed to indicate that "poorest" readers failed to check
the best response, for the reason just given.

APPENDIXES 331

The cause of the lowest V in each test seemed apparent after critical analy-
sis. In Form I, the lowest V> .428, was found for Item Number 6, a selection
from the Bible. It appeared from the comparative comprehension of "best"
and "poorest" readers that familiarity with the passage had tended to reduce
its discriminating power. The lowest ^in Form a, .297, was found for Item
Number 8. Examination of the responses to this item indicated that the unin-
tentional introduction of a "catch" word in the poorest response invalidated
the item. The poorest response was "A little linseed oil gives a gloss to the
feathers of a canary/' The "catch" word 0/7 was ignored by all readers, with
the result that "'best'* readers tended to fail in seeing that the response was
wrong about as often as ''poorest" readers. With the exception of these
items, the items in both tests showed generally rather high discriminating
power.

ARE THE TESTS RELIABLE INSTRUMENTS FOR
MEASURING ADULT COMPREHENSION?

The second most important fact which must be known about a test is the
degree of reliability it possesses. Before producing evidence to show how well
the adult reading tests measure whatever they do measure, we will briefly sum-
marize the means used in the development of the tests to provide for re-
liability.

Objectivity of scoring. If a test is perfectly objective, that is, if answers
which are given credit are sharply defined in a key, and only those answers
given credit, one factor of influence on reliability may be eliminated. In the
present investigation, it should be recalled, all answers were given some credit,
on the assumption that particular numerical values assigned to test responses
provide a satisfactory scale for measuring the degree to which readers have
attained the objectives measured by the tests. In order to reduce the influence
of personal judgment in scoring, composite evaluations were secured for each
response from a number of trained judges. That a high reliability of evalua-
tion was obtained by these judges has been shown by coefficients of reliability
ranging from ,994 to .970. By the use of a scoring stencil, on which each re-
sponse was marked with its estimated value, subjectivity of scoring was re-
duced to a minimum.

Character of sampling included in the test items. Other things being equal,
the more extensive the sampling the more reliable the test. In an attempt to
increase the reliability of the measure, sample paragraphs were taken from a
wide sampling of books, magazines, and newspapers, and as many of these
were included in each test as could be reasonably answered by adults of limited
reading ability in a testing period of one hour,

The reliability of the tests. Data concerning the reliability of the tests de-
veloped in this investigation are presented in Table LXXXI. These data were
obtained from the scores of 756 adults, who took both forms of the test. The
correlations between Forms i and 2 varied from .779 for evening-school groups
to .481 for colored groups in the A and M Junior High School. The mean
coefficient of reliability was not calculated, nor was the coefficient of reliability

33*

WHAT MAKES A BOOK READABLE

for the entire 756 subjects, since it may be materially affected by the hetero-
geneity of the total population from which the data were collected.

An interpretation of coefficients of reliability without recognition of the
range of talent represented in the data from which they were computed is apt
to be misleading. This consequence was avoided by measuring the variability
of the differences between the obtained scores and the corresponding theoreti-

TABLE LXXXI

RELIABILITY COEFFICIENTS, STANDARD DEVIATIONS, PROBABLE ERRORS OF

SCORES, AND RATIOS BETWEEN THE PROBABLE ERRORS AND STANDARD

DEVIATIONS OF SCORES ON ADULT READING TESTS*

Group

r P.E.

(Tx

<T2

P.-E.Score

wwx

I. 2. 7

770 + O72

22 7C

17. 7O

6.41

778+ O28

22. 1C

18.40

6-4<

15,16

777+ Od.8

j
17. 7C

i6.ic

v -*h3

C.86

.7OO

A / * /j
21 .70

*J
19.63

7.63

Jj
17

6oi oci

22 Q<

22. 6c

8.cc

2O, 21

.666+ .041

II 1C

1C. 1C

u j j

.7Q

17 OO

I2.QO

c 23

.70

(v 10. 1 1

co6 + oci

I8.IC

12 Q2

6.66

.4.7

.567

10. ^O

1 o fg

7. 7O

.44.

18* IQ

. C7Q + .oco

y x

17. oC

1C. 40

7-<7

,46

K.4O

12. 2O

6.46

.47

6 7, 8

.481 .O7C

14. 4.C

6 66

* Coefficients of correlation and standard deviations are quoted from Cleary, op. cit., p. 84.

cal true scores. The median deviation of the differences or the probable error
of the scores was obtained by the formula: 31

in which - represents the average standard deviations for Forms I and 2,

and TI/ represents the correlation of Forms i and 2.

Table LXXXI shows the probable error of the scores of different groups
used in finding the correlation coefficients which are shown in the same table.
The first probable error of 6.41 means that there is an even chance that a score
made by an evening-school student on the test is within 6.41 points of his true
score. For example, if an individual made a score of 68.9, the chances are even

a 1 G. M. Ruch, The Objective or New-Type Examination (Chicago: Scott, Foresman
& Co., 1 929), p. 430.

APPENDIXES 333

that the average of his scores obtained from an infinite number of test forms
would lie between 62.5 and 75.3. Since the highest scores obtainable on Forms
i and 2 are, respectively, 92.2 and 91.4, the probable errors do not seem dis-
proportionately large. The ratio of the probable errors to the standard devia-
tions is between one-third and two-fifths for most groups.

Since no data from other adult tests are available for comparative purposes,
no conclusions can be drawn with respect to the relative reliability of the adult
reading tests developed in this study. That the reliability of the tests is hardly
high enough to be considered entirely satisfactory must be granted, since
they fall within the category defined by Ruch as "rather low." 32

3* Ibid., p. 434.

APPENDIX D

As stated in chapter iv, the identification of elements of difficulty depended
on the correlation between the occurrence of the elements in a series of test
items and the average reading score "made by adult readers on those items.
The method of tabulating individual scores and of obtaining average scores is
described in the following paragraphs.

TABULATION OF SCORES

Two procedures were followed in tabulating scores* One provided a com-
pact record of each subject's scores for both test forms; the other afforded a

Form

</ o

Item No.
15 omit-
ted from
Form 2.

3.8

3-3

3-8

3.9

3-9

S. Mor.
N. Sch.

D
Cles

M-

iner

Z .9

|l.S

3.0

3.9

4-o

99* 4,3

3 1 "^

1 M

1 1

2.4

3-0

3-7

6.7

o| |V|

o|v|

v| |

3.7

3-7

4.0

2.7

58.0

1 ! 1

2.O

FIG. 32. Tabulating card for recording individual test scores.

Data recorded at the left are read down: Column I reads as follows: Sterling Mor-
ton Evening School, Student Number 21, white, male, age 35, last grade in school, 8,
attendance at nignt school, 3 years, 3 months. Column II reads as follows: average read-
ing per day, 45 minutes; occupation, dry cleaner; Monroe test, rate 99 J, "B" score 4.3;
comprehension 7, "B" score 6.7; total score on Form 2, 58.0.

Data recorded in other columns are read across as follows: score on first item, 3.8;
on second item, 3.3; on third item, 3.8; and so on. Blank spaces indicate items not at-
tempted.

means of determining average scores on all items of a test for a particular group.
The first procedure involved the use of tabulating cards, one for each person
for each test. These cards were made of tag-board, f^XS", for convenient
filing. Data from the test booklets were entered on the card according to
definite spatial arrangement. Personal information furnished by the subject on
the cover-page of his test booklet was entered on the left side of the card, and a

334

APPENDIXES

335

duplication of his checked responses for each item was indicated in the spaces
provided in the main body of the card.

Figure 32 is^a copy of one subject's record for Form 2. In making such a
record, the subject's responses were first copied directly from his test booklet,
by marking " V " and "o" in the spaces corresponding to the response order in
the booklet. All test booklets could then be filed, and the cards substituted for
further use. A stencil was designed to fit over the card, showing the value of

TABLE LXXXII

CRITERIA OF DIFFICULTY FOR ITEMS ON FORM i, REPRESENTED BY THE AVERAGE
SCORE OF ALL READERS, "BEST" READERS, AND "POOREST" READERS

TEST
ITEM

AVERAGE READING
SCORE

TEST
ITEM

AVERAGE READING SCORE

All
Read-
ers

"Best"
Read-
ers

"Poor-
est"
Read-
ers

All
Readers

"Best"
Readers

"Poorest"

Readers

2.6

2 -3

2,1
2.6

a -5
3-

2.6

2.2
2.0
2. 4
2.1

i-9

3-7
3-o
3-3
3-a
3-a
3-7
3-8

3-a
3,o
3-8
3-3

3-4

i.i

T -5
0-5
i-S
i-5

2.0

i-3

1,0

0.6

0.4
0.7

0.2

2.4
a-3
M
i-9

a- 5
i-9
2.4

2.1

*-9

2.0

i-9

*-7

3-4
3-i

3-4

2.8

3-6
3-4
3- a
3-3
3- 2
3-5
3- 1
3-2

i.i

I.O
1.2

o-7

I.O
O.I

0.9

-7
0.6

o-5
0.7
o. 4

^ .

I:::::'

2T .

Id. .

Ranee.

3.0-1.7
2.2 .070
5io .050

3.8-2,8
3.3+. 066
-480 .047

2.00.1

o.9 .058

.424* .041

Mean .

S.D

each response in all paragraphs. It was then possible to make rapid calcula-
tion of the score for each test item by remembering the formula: o minus V =*
score. The score for each item was then entered in the proper space, as shown
in Figure 32, By adding algebraically all scores on the card, the total test
score was obtained and entered at the bottom of Column 2. Although consid-
able labor was expended on the card tabulations, it appeared to be justified by
the convenience of compact records.

COMPUTING AVERAGE READING SCORES

The purpose of finding average scores was to obtain a criterion of difficulty
for each test item that might be used in determining elements of difficulty

33 6

WHAT MAKES A BOOK READABLE

within the item. According to the plan described in chapter iv, a criterion of
difficulty was obtained for each item for all readers, for "best" readers and for
"poorest" readers. "Best" readers were those individuals whose combined
scores on Forms I and 2 of the Adult Reading Test were above the third quar-
tile. This group numbered 191. "Poorest" readers were those whose combined
scores on the two test forms fell below the first quartile. One hundred ninety
individuals were identified as "poorest" readers among the twenty-one groups
tested.

TABLE LXXXIII

CRITERIA OF DIFFICULTY FOR ITEMS ON FORM 2, REPRESENTED BY THE AVERAGE
SCORE OF ALL READERS, "BEST" READERS, AND "POOREST" READERS

TEST
ITEM

AVERAGE READING
SCORE

TEST
ITEM

AVERAGE READING SCORE

All
Read-
ers

"Best"
Read-
ers

"Poor-
est"
Read-
ers

All
Readers

"Best"
Readers

"Poorest"
Readers

2.9

2.6

2.4

2.6
2.8
2.6

i-3

0.8

2.6

3-
i-3
a -5

3-5
3-5
3-2

3- a
3-5
3-*
*$

1.2

3-|
3-6

2.8

3-2

2.1
1.2

i-7
1.6

M
'-7
-5

O.I

i.i

2.0
0.2

*-3

TO . ...

a-3
i-S

2.6

1.8
i-9

2.2

i-5
i-3
0.9
0.7
1.8
1.8

3-3
2-3
3-3
3-o
3-o
3-7
3-i
3-i
2.7
1.6
3-a
3-3

I.O

0.9

o-S
-5

0.2

o-3

O.I
~O.I
0.2

-3

0.7

^ . .

4.
5

7 ...

IQ.

2O .

12 ...

Ranee

2.9-0.7
2.o.097
.707+. 069

3-7- 1 - 2
3--95
693 .067

2.1 0.2

o.9.o87
.6331.062

Mean ,

SJD

In finding the three sets of average scores, three sets of master-sheets were
prepared. One recorded individual scores on each test item for each of 759
readers. The other two contained individual scores on each item for each
"best" and "poorest" reader, respectively. Average reading scores on each
item were then calculated for the three groups, as shown in Tables LXXXII,
and LXXXIII. These scores are the criteria of difficulty used in identifying
elements of difficulty shown in chapter iv.

APPENDIX E

In Table LXXXIV are listed the coefficients of correlation obtained by
correlating the forty-four elements of difficulty with the average reading score
of three classes of readers and with each other.

337

:M 1 *

r, ^Q P P EJ

8t-3-3-3

I S < 3 < 3o-5|S-S-og~S J a&g'S&^'S'S'S

ft 1 " ^ f SJ^

g 6

TABLI

CORRELATIONS OF ELEME

AVERAGE REAPING SCORES

Average
Reading
Scone

Average Reading Scan

Element* of

9538

8&17

4786

Elements

18ZS

0514

1678

0695

0038

0933

1563

-W53

1977

-1530

-1774

1777

-Z349

-1(95

2853

0397

05&5

I7C5

1463

0452

1268

1637

<X>i9

0316

0236

Z6IZ

1753

2323

0616

1251

;040I

0917

200

-tut

-1733

2323

0616

ies\

0401

0917

-2100

1000

$196

2714

5291

H8Z

1203

-1130

0511

2180

69f2

-5134.

-2181

5497

-(734

0440

2497

.0944

2855

6Z5

6265

6493

-3804

-iao6

4414

0055

Oltt)

0140

2404

2321

4393

5593

.4244

-4J64

-1911

5454

-IZ33

1227

2158

0876

2104

6631

663 >

9629

J5764

5983

4326

.1660

4853

2394

0881

1275

2062

2940

.5022

5022

8360

8329

-3989

-0186

-i9za

1688

1870

2518

1518

0303

1350

1420

2740

%36

'2466

1775

335?

47Z4

-3796

0774

4275

-1145

0207

1596

.1564

2479

M3IO

4310

7429

8009

3488

.7336

8613

OZ83

-5217

431 \

4488

Z500

1080

2612

1679

0988

-0805

0805

3978

,4853

07S8

3903

4456

0691

4621

-47ZO

-2868

4S7E

2668

r.fl2G

245Z

167?

:2I33

-.2594

2594

5732

1707

5580

6572

1050

6871

9156

369?

.3888

2382

I960

0206

-212&

-H85

1157

0242

242

2226

2395

0433

-1671

2472

JQS61

-3089

-7749

-6915

.I32B

0057

2553

Z076

2748

-1603

H567

1930

.12)4

IZtt

1892

-2678

.1062

-17 Id

1887

JOS92

1527

-3497

-3051

1012

Z046

2485

137B

0517

-1695

0104

3524

J0303

3764

-3764

2&43

1826

-.273)

-2524

.2136

-1701

1455

I&94

0468

-2059

-5211

4623

3603

4190

Z949

1219

2360

,0924

0209

.1207

-,IZ07

4207

*635

0885

-3618

.4233

0483

4345

-5447

-7865

8241

4919

1945

097)

0794

ZIIZ

05)6

01%

0081

0924

1453

-0734

0734

-1472

0639

.36? 1

1150

081?

0305

0829

2361

2420

-333?

0620

02&I

-2329

08&4

37C7

0856

0744

3413

0079

-34J6

2647

3571

.4126

0539

44)6

2834

0603

.3B0I

I73Z

268)

0336

-1700

OJtt

1588

0384

2996

0573

.0345

-1547

-1146

11)9

2150

-2150

2620

^794

-Z033

-27W

.1365

07 18

-1147

-1353

-1712

0590

0245

2225

3447

0625

3655

-ZZ48

-1070

.2809

3246

0082

D&I9

0619

5094

506)

1075

5153

4992

1747

4288

4?90

4550

-Z6&5

-529?

IOO

-2833

0500

Z7\l

-2551

-.1241

Z3%

2941

0507

0166

-4307

4583

0352

4325

4674

.1545

4024

4052

4399

-2547

-.5407

2179

I94Z

2717

1105

0479

-0391

2693

0107

O244

2196

ZI56

2630

-2912

0741

-I8

2518

0497

-29T&

-3349

-3489

9605

208>

0789

3Z53

1500

3070

3448

0344

0517

-2431

1575

!&&

2186

>948

-4541

-3275

-3799

4003

2?IZ

-.3404

-3897

4021

1549

.2077

.1738

4756

3468

4502

238S

0035

2S43

-149?

0932

4090

4080

C.II6

-7133

-1897

-5385

5931

1623

^7SI

tf396

6599

4273

.7682

0647

P 1

4311

3364,

3711

2978

1407

226?

J0859

0191

0953

3745

-4371

d4!4

-.3154

4107

602

4144

-,8101

7519

7771

5092

:I967

Q9J2

168)

0604

iaw

0040

7521

0327

0457

0139

J0796

-0184

0629

04 f4

1262

0272

10739

0638

J0203

.1559

.0274

r .

5110

921

9180

1(233

IZ87

0258

0727

1796

6520

3517

-8475

5850

8896

60%

3651

7021

9542

-5184

2080

2015

2447

-aw

0753

3666

0661

0336

1358

2039

1931

4197

5074

-4044

8959

C027

3721

.IZ27

3536

0784

.I&73

0234

,932

:23

4740

3016

4070

2033

0564

286fc

(290

1231

2691

5958

6305

.2180

.6035

5796

1356

5770

7943

096

5498

-.3340

0876

I 1

^13

4800

-4059

-2230

0277

1069

2755

1159

1197

2S74

.4137

0849

.2336

*!04!

0257

4517

8506

0d4

7212

-1677

0696

4020

Z2&9

4086

0657

0058

2004

0749

.1677

-4694

4694

7190

.6774

.3784

6675

685Z

1732.

6655

4361

,5547

Z677

2100

T0942

-2288

-35*4

0932

0936

ZI03

0356

-1047

1551

240

2419

1138

0841

2323

1123

1435

0022.

.1870

2818

.2778

4643

C736

"J4777

.0537

1140

OZ3I

J 191

2013

.0214

.3625

0323

3378

1105

-0478

2511

-.1067

J073Z

6?Z

0501

.5113

3964

502Z

8937

.1*01

2764

J09W

0013

-0853

-I40&

,1564

-I8SI

0421

0356

0443

,0054

0194

017?

0816

0494

*740

-3707

T8&9

ntttt

,1440

.1414

1942

0655

0337

17&0

1024

.2413

0444

3762

2880

-2439

-.2744

-2464

2432

J0823

-.2328

T043I

-,)Z60

.0826

5203

&ZO

0087

0497

0750

OOZO

2123

.1215

O599

3613

1520

-.0898

.2944

-1371

J0765

0180

08S7

3276

2041

^746

.5711

7315

:32I4

:J 14-1

786

1241

0544

.Z807

0434

-1484

3161

3)61

-5926

5652

2250

.6115

.5199

1347

5126

5497

006

-3169

-3308

.0424

0940

16)

0185

,(535

0132

0643

0547

.1963

2146

2)46

r|4)7

-2641

372C

3267

.4)26

1264

1312

0366

-.2034

0295

1)55

.1)67

.0282

2341

O06Z

0000

JOS85

-1877

.1176

0932

C932

067)

10523

JOZI4

052F

JOC7)

202)

0461

0&I4

.1188

451

3335

0283

S 1

JO))

-\yn

:177Z

1075

tan

-.1219

.0796

0069

0231

.0231

1309

-0)87

2698

.1950

165)

.1302

1353

.1441

.Oftl!

-235?

,2871

0291

LLXXXIV

S OF EXPRESSION WITH

,NP WITH EACH OTHER

ras*i

Mean

F 1

G 1

I 1

J 1

K 1

M 1

P 1

0.'

t II

I 17

3.17

038

087

057

373

f,87

423

146

215

154

406

173

578

230

5492

503

4506

503

7406

807

963

621

W.75

506

2612

9.26

7092

6.99

18.23

339

IO.H7

6.26

19.24

709

2731

II.&5

292

2.00

2.4Z

1.51

0.60

0.64

6.50

2.61

0.94

-,Z271

0.61

1.32

0213

a?55

163

1.30

0300

4433

2828

1176

356

-1112

3344

0824

9326

13.21

345

2472

2554

0633

3053

-3465

2.52

460

1446

0461

.1759

1992

-1407

2466

642

469

0634

O042

J0672

4695

4550

5837

.5556

954

5.96

-I62G

0664

0260

5007

5300

5267

.2966

964

675

3.03

Z098

,2164

0048

0129

0116

-0662

1973

0715

363

12.65

3.90

-.1674

2971

Z408

4672

.2169

4308

5934

4111

3639

636!

0,31

Z445

1275

1396

0689

0191

0564

-2407

1428

0737

Z996

4077

7373

5.72

0666

3310

iX)33

4464

.4625

3001

4256

6323

fitte

0930

5361

669

49.00

9.23

2000

0464

1735.

ZBI1

.2268

2730

-3341

sat

6663

0021

2755

0598

5516

12.04

A 12

.1195

.2642

0391

4052

406?

3237

3397

339

4390

1112

7079

3097

A782

3801

3>.56

747

1724

J0252

1990

J973

1255

0667

1639

2336

0293

1436

2123

1666

.3664

1159

35.3?

0961

0752

0955

2561

3041

1578

0369

1191

358?

1122

1469

1933

3124

3292

0116

J527

12,00

:363)

M63

1166

0742

.1130

0905

JO 133

05M

2731

Z192

4660

2390

5365

0032

5763

3005

43.77

20.16

0077

07)9

-1155

0906

0493

1303

1636

2664

028>

1302

3242

ZI99

1525

0221

2604

5964

5460

1690

1.12

t.07

0728

0696

.1296

913

Z536

-0739

0163

-oiao

2748

.220)

2271

0906

Z775

0754

4637

7761

3530

6651

2417

6.16

0376

0605

4114

4502

2282

iMI9

5)00

4476

1354

5212

2968

B99S

1952

J6BI

0226

.21 tO

JK66

-I6U

0186

37.21

G42

1327

0951

J06IG

I7*

;29

1749

0629

1579

rtfSS

0722

5524

346

0660

1474

03)1

0&66

1455

1655

0311

[439

1543

t 17

1,25

0362

0411

j0557

.1056

4J690

0157

1924

1575

2105

*565

1660

019

0225

(155

0516

2622

0715

548

.1662

.1191

*4W

too

1.15

101

1627

.1306

166t

1444

0696

0)13

1021

r,K35

II5

1625

265

1616

1961

4659

4796

0126

2661

;1924

1979

((76

-,IIW

0000

163

131

APPENDIX F

Table LXXXV shows the predicted difficulty of 350 books, obtained by
the regression equation for variables 1.25678, described on page 138 of chap-
ter iv. Difficulty is expressed in terms of the average comprehension score
that adults of limited reading ability would probably make if tested on the
material. As in the case of true scores obtained by actual testing, high pre-
dicted scores indicate selections easily comprehended, and low scores, selec-
tions difficult to comprehend.

A graphical presentation of the books listed here appears in Figure 23,
chapter vi.

TABLE LXXXV

PREDICTED DIFFICULTY or 350 BOOKS

Predicted

No. Book Index of

Difficulty

1. Abraham, Robert Morrison. Winter Night's Entertainments; a
Book of Pastimes for Everybody. Dutton, 1933 1 . 1 1

2. Adams, H. C. Travelers 9 Tales. Boni, 1927 54

3. Adams, James Truslow. The Epic of America. Little, 1931 ... .15

4. Adams, Randolph G. The Gateway to American History. Little,

I9 2 7 87

5. Adams, Randolph G. Pilgrims, Indians and Patriots. Little,

1928 : 87

6. Aitchison, Allison, and Uttley, Marguerite. Across Seven Seas

to Seven Continents. Bobbs, 1925 93

7. Akeley, Carl E. In Brightest Africa. Garden City, 1923 82

8. Alcott, Louisa M. Little Women. Little, 1915 42.

9. Aldrich, Bess Streeter. A Lantern in Her Hand. Apple ton, 1928 .86

10. Allen, Frederick Lewis. Only Yesterday. Harper, 1931 06

11. Anderson, Robert Gordon. Those Quarrelsome Bonapartes*
Century, 1927 52

12. Andrews, Margaret Lockwood. The Complete Book of Parties.
Funk, 1932 . 88

13. Andrews, Mary R. S. The Perfect Tribute. Scribner, 1912 74

14. Anthony, Katherine. Queen Elizabeth* Knopf, 1929 86

15. Antin, Mary. At School in the Promised Land. Hough ton, 1911 .86

1 6. Antin, Mary. The Promised Land. Houghton, 1912 86

17. Arliss, George. Up the Years from Bloomsbury; an Autobiog-
raphy. Little, 1928 46

1 8. Auslander, Joseph, and Hill, F. E. The Winged Horse. Double-
day, 1928 92

339

340 WHAT MAKES A BOOK READABLE

TABLE LXXXV Continued

Predicted

No. Book Index of

Difficulty

19. Austen, Jane. Pride and Prejudice. Grosset, 1931 46

20. Austin, Mary. The Land of Journey s Ending. Century, 1924. . .23

21. Bailey, Temple. Contrary Mary. Grosset, 1914 1 .03

22. Balzac, Honor6 de. P$re Goriot. Little, 1931 38

23. Banks, Helen Ward. The Story of Mexico. Stokes, 1926 40

24. Barnes, Gerald. Swimming and Diving. Scribner, 1922 55

25. Barnes, James. Drake and His Yeomen. Macmillan, 1899 69

26. Barnes, Parker T. House Plants, and How To Grow Them.
Doubleday, 1909 88

27. Barryrnore, John. Confessions of an Actor. Bobbs, 1926 62

28. Harriett, Robert A. The Log of Bob Bartlett. Putnam, 1928 ... .86

29. Barton, Bruce. What Can a Man Believe? Bobbs, 1927 86

30. Bates, Sylvia Chatfield. Moby Dick (adapted). Scribner, 1928 .75

31. Bauer, M,, and Peyser, E. How Music Grew. Putnam, 1925. . .57

32. Beach, Rex. The Silver Horde. Burt, 1909 82

33. Beals, Carleton. Brimstone and Chili; a Book of Personal Ex-
periences in the Southwest and in Mexico. Knopf, 1927 41

34. Beard, Annie E. S. Our Foreign-Born Citizens. Crowell, 1922. . .65

35. Becker, May Lamberton. Adventures in Reading. Stokes, 1927 ,62

36. Becker, May Lamberton. Books as Windows. Stokes, 1929. . . .57

37. Beraud, Henri. Twehe Portraits of the French Revolution. Little,

1928 53

38. Berge, Victor, and Lanier, Henry. Pearl Diver; Adventuring

over and under Southern Seas. Doubleday, 1930 52

39. Birkhead, Alice. The Story of the French Revolution. Crowell,

19^3 65

40. Bok, Edward. The Americanization of Edward Bok; an Auto-
biography. Scribner, 1921 83

41. Borup, George. A Tenderfoot with Peary. Stokes, 1911 63

42. Bowers, Claude G. The Tragic Era. Houghton, 1929 19

43. Bowrnan, Charles Ellis, and Percy, A. L. Fundamentals of Book-
keeping and Business. American Book, 1926 i .07

44. Boswell, James. The Life of Samuel Johnson. Modern Library,

1931 23

45. Boyd, James* Marching On. Scribner, 1927 74

46. Bridgman, George. Constructive Anatomy. Bridgman, 1920. . , 1.08

47. Brigham, Louise. Box Furniture. Century, 1910 1.24

48. Bromfield, Louis. The Green Bay Tree. Stokes, 1924 69

49. Browne, Lewis. Stranger Than Fiction. Macmillan, 1925 67

50. Bryan, George S. Edison; the Man and His Work. Knopf, 1926 .54

51. Buck, Pearl S. The Good Earth. Day, 1931 81

52. Bullen, Frank T. The Cruise of the Cachalot. International
Book, 1899 24

APPENDIXES 341

TABLE LXKX.V Continued

Predicted

No. Book Index of

Difficulty

53. Burnham, Smith. Our Beginnings in Europe and America, Win-
ston, 1918 90

54. Burns, Walter Noble. The Saga of Billy the Kid. Doubleday,

1926 58

55. Burns, Walter Noble. Tombstone. Doubleday, 1927 69

56. Burt, Emily Rose. Planning Your Party. Harper, 1927 86

57. Byrd, Richard E. Skyward. Cornwall, 1928 49

58. Canfield, Dorothy. The Deepening Stream. Harcourt, 1930. . . .62

59. Cantor, Eddie. My Life Is in Your Hands. Harper, 1928 63

60. Carnegie, Dale. Lincoln the Unknown. Century, 1932 76

61 . Carroll, Gladys Hasty. As the Earth Turns. Macmillan, 1933 . . .86

62. Cary, Katharine T., and Merrell, Nellie D. Arranging Flowers
throughout the Year. Dodd, 1933 54

63. Casey, Robert J. The Lost Kingdom of Burgundy. Century,

!9*3 59

64. Casey, Robert J. Four Faces of Siva. Bobbs, 1929 56

65. Gather, Willa Sibert. The Song of the Lark. Houghton, 1915. . .82

66. Cendrars, Blaise. Suiters Gold. Harper, 1926 73

67. Center, Stella Stewart. The Worker and His Work. Lippincott,

1926 48

68. Chambers, Robert W. Cardigan. Harper, 1901 31

69. Chase, Stuart. A New Deal. Macmillan, 1932 29

70. Chekhov, Anton. The Lady with the Dog> and Other Stories.
Macmillan, 1928 68

71. Churchill, Winston. The Crisis. Macmillan, 1901 87

72. Churchill, Winston. Richard Carvel. Grosset, 1914 67

73. Clark, Barrett H. How To Produce Amateur Plays. Little,

1921 72

74. Clark, Keith. The Spell of Spain. Page, 1914 65

75. Claudy, C H. The First Book of Photography. McBride, 1918 .76

76. Coffin, Charles Carleton. The Boys of '76. Harper, 1876 82

77. Coffin, Charles H. How To Study Pictures. Century, 1918 43

78. Collins, A. Frederick. The Book of the Microscope. Appleton,

i9 2 3 7 1

79. Colum, Padraic. Cross-Roads in Ireland. Macmillan, 1930. . . .94

80. Conner, Ralph. The Sky Pilot. Revell, 1899 84

81. Connolly, James B. The Book of the Gloucester Fishermen. Day,

1927 84

82. Cooper, Courtney Ryley. Go North, Young Man! Little, 1929 .32

83. Davies, Blodwen. Romantic Quebec. Dodd, 1932 56

84. Davis, William Stearns. A Friend of Caesar. Macmillan, 1922 .53

85. Davis, William Stearns. Life in Elizabethan Days. Harper,

193 33

342 WHAT MAKES A BOOK READABLE

TABLE LXXXV Continued

Predicted

No. Book Index of

Difficulty

86. Daudet, Alphonse. Tartarin of Tarascon. Little, 1927 44

87. Deeping, Warwick. Sorrett and Son. Knopf, 1926 71

88. D'Esque, Jean Louis. A Count in theFo'csle. Brentano's, 1932 . 87

89. Defoe, Daniel. Robinson Crusoe. Scott, Foresman, 1913. (Un-
altered from sixth edition, 1722.) . 26

90. DeKruif, Paul Henry. Microbe Hunters. Harcourt, 1926 70

91. Dickens, Charles. David Copperfield (condensed by R. Graves).
Harcourt, 1934 85

92. Dickens, Charles. The Personal History of David Copperfield.
Burt, n.d 16

93. Dhnnet, Ernest. The Art of Thinking. Simon, 1928 51

94. Dix, Mark H. An American Business Adventure; the Story of
Henry H. Dix. Harper, 1928 59

95. Doyle, A. Conan. The White Company. Harper, 1894 45

96. DuChaillu, Paul B. Country of the Dwarfs. Harper, 1928 92

97. Duguid, Julian. Green Hell. Century, 1931 40

98. Dumas, Alexandre. The Three Musketeers. McNally, 1923 69

99. Earhart, Amelia. The Fun of It; Random Records of My Own
Flying and of Women in Aviation. Putnam, 1932 62

100. Eaton, Jeanette. A Daughter of the Seine. Harper, 1929 55

101. Eaton, Jeanette. Young Lafayette. Houghton, 1932 45

102. Eddy, Clyde. Down the Worlds Most Dangerous River. Stokes,

J 9 2 9 56

103. Edwards, Isabel M. Glove-Making. Pitman, 1929 78

104. Eells, Elsie Spicer. South Americas Story. McBride, 1931 64

105. Ekrem, Selma. Unveiled; the Autobiography of a Turkish Girl.
Washburn, 1930 86

1 06. Eliot, George. Silas Marner (edited by A. E. Hancock).
Scott, Foresman, 1899 13

107. Ellis, Anne. tl> Plain Anne Ellis"; More about the Life of an Ordi-
nary Woman. Houghton, 1931 87

108. Ellsberg, Edward, On the Bottom. Cornwall Press, 1928 71

109. Erskine, John. The Delight of Great Books. Bobbs, 1928 55

1 10. Fairbank, Janet Ayer. The Bright Land. Houghton, 1932 74

in. Finger, Charles J. Courageous Companions. Longmans, 1929. . .86

112. Finger, Charles J. David Livingston. Doubleday, 1927 69

113. Finger, Charles J. Footloose in the West; Being the Account of a
Journey to Colorado and California and Other Western States.
Morrow, 1932 58

114. Ferber, Edna. Cimarron* Grosset, 1929 78

115. Ferber, Edna. Fanny Herself. Stokes, 1917 82

116. Ferber, Edna* Show Boat. Grosset, 1926 79 j

APPENDIXES 343

TABLE LXXKV Continued

Predicted

No. Book Index of

Difficulty

117. Ferris, Helen, and Moore, Virginia. Girls Who Did. Button,

1927 I. oo

1 1 8. Fierro Blanco, Antonio de. The Journey of the Flame. Hough-
ton, 1933 40

119. Fleming, Peter. Brazilian Adventure. Scribner, 1934 55

no. Forster, Edward M. A Passage to India. Harcourt, 1924 61

121. Fosdick, Harry Emerson. Twelve Tests of Character. Associa-
tion Press, 1928 68

1 22. France, Anatole. The Crime of Sylvestre Eonnard. Harper, 1 890 . 67

123. Frank, Harry A. A Vagabond Journey around the World. Cen-
tury, 1920 40

124. Frank, Leonhard. Carl and Anna (translated into Basic Eng-
lish by L. W. Lockhart). Trubner, 1930 99

125. French, George W. Photography for the Amateur* Folk, 1922. . .62

126. Friese, John F. Farm Blacksmithing. Manual Arts Press, 1921 i . 13

127. Gardener, Elmer Ellsworth. Better Typewriting. Prentice-Hall,

1931 ; 75

128. Garnett, David. Pocahontas. Harcourt, 1933 67

129. Garretson, Edith May. Home and Health in a New Land. Scrib-

ner, 1927 i . 57

130. Gaye, Phoebe Fen wick. Vivandidre! Liveright, 1929 66

131. Gibbons, John. Afoot in Italy. Dutton, 1932 68

132. Gibbs, Philip. Since Then; the Disturbing Story of the World at
Peace. Harper, 1930 39

133. Gide, Charles. First Principles of Political Economy. Harrap,

19*2 ; 75

134. Goldberger, Henry H. America for Coming Citizens. Scribner,

1922 90

135. Goldsmith, Oliver. The Vicar of Wakefield. Macmillan, 1922. . .25

136. Goldstein, H., and Goldstein, V. Art in Everyday Life. Mac-
millan, 1929 90

137. Gordon, Jan and Cora. Two Vagabonds in Spain. McBride,

19^3 7

138. Gras, Felix. The Reds of the Midi; an Episode of the French Revo-
lution (translated by C. A. Janvier). Appleton, 1923 71

139. Grey, Zane. Riders of the Purple Sage. Grosset, 1912 92

140. Greene, Anne Bosworth. Lighthearted Journey. Century, 1930 .35

141. Grenfell, Wilfred Thomason. Tales of the Labrador. Houghton,

1916 66

142. Groves, Ernest R. and Gladys H. Wholesome Childhood.
Houghton, 1924. 71

143. Hammett, Charles Edward. Major Sport Fundamentals. Scrib-

ner, 1927 , 7

344 WHAT MAKES A BOOK READABLE

TABLE ISS3X Continued

Predicted

No. Book Index of

Difficulty

144. Hamsun, Knut. Growth of the Soil. Knopf, 1930 91

145. Harris, Stanley, Baseball How To Play It. Stokes, 1925 97

146. Hartmanj Gertrude. These United States and How They Came

To Be. Macmillan, 1932 78

147. Hathaway, Esse V. The Book of American Presidents. Whit-
tlesey House, McGraw, 1931 80

148. Hawks, Frank. Speed. Putnam, 1931 50

149. Hemon, Louis. Maria Chap delaine. Grosset, 1924 31

150. Henderson, Rose. Little Journeys in America. Southern, 1923 .54

151. Henry, O. The Four Million. Doubleday, 1922 70

152. Heary, Robert Selph. Story of the Confederacy. Garden City,

W 43

153. Herdman, Marie Louise. History of the United States. Stokes,

1916 70

154. Hervey, Harry. King Cobra; an Autobiography of Travel in
French Indo-China. Cosmopolitan, 1927 38

155. Hewitt, Edward Ringwood. Telling on the Trout. Scribner, 1926 .72

156. Heyward, DuBose. Peter Ashley. Farrar, 1932 73

157. Heyward, DuBose. Mamba's Daughters. Doubleday, 1929 65

158. Hibben, Thomas. The Carpenter s Tool Chest. Lippincott, 1933 .98

159. Hildebrand, Arthur Sturges. Magellan. Harcourt, 1924 62

160. Hill, Janet McKenzie. The Up-to-Date Waitress. Little, 1922. . ,74

161. Hindus, Maurice. Red Bread. Smith, 1931 60

162. Hodgins, Eric, and Magoun, F. A. Sky High; the Story of Avia-
tion. Little, 1929 58

163. Hogue, Wayman. Back Yonder; an Ozark Chronicle. Balch,

*93 2 97

164. Holdridge, Desmond. Pindorama. Minton, 1933 68

165. Hough, Emerson. The Covered Wagon. Grosset, 1922 76

166. Hubbell, Jay B. The Enjoyment of Literature. Macmillan, 1929 .65

167. Huberman, Leo. 'We, the People" Harper, 1932 68

168. Hueston, Ethel. Coasting Down East. Dodd, 1924 27

169. Hulit, Leonard. The Salt-Water Angler. Appleton, 1924 61

170. Ilin, M. New Russia's Primer; the Story of the Five-Year Plan
(translated from the Russian by George S. Counts and Nucia P.
Lodge). Houghton, 1931 97

171. Ilin, M. What Time Is It? Lapshin, 1932 i .04

172. Irving, Washington. The Bold Dragoon. Knopf, 1930 56

173. Irwin, Margaret. Royal Flush; the Story of Minette. Harcourt,

193* 57

174. JarTe, Bernard. Crucibles. Simon, 1930 69

175. James, Bessie R., and James, Marquis. Six Foot Six; the Heroic
Story of Sam Houston. Bobbs, 1931 83

APPENDIXES 345

TABLE LXXXV Continued

Predicted

No. Book Index of

Difficulty

176. Johnson, Charles H. L. Famous American Athletes of Today.
Page, 1928 53

177. Johnson, Martin E. Safari; a Saga of the African Blue. Put-
nam, 1928 64

178. Johnston, Marjorie. Domination. Appleton, 1930 38

179. Judd, Alfred. The Conquest of the Poles; and Modern Adventures

in the World of Ice. Nelson, 1924 25

1 80. Kang, Younghill. The Grass Roof. Scribner, 1931 92

181. Keller, Helen. The Story of My Life (School Edition). Hough-
ton, 1904 80

182. Kellock, Harold. Houdini; His Life Story. Harcourt, 1928. .. .55

183. Kirkpatrick, Frank Home. Public Speaking. Doran, 1923. ... i .22

1 84. Klein, Paul E. Shoe Repairing. Bruce, 1926 88

185. Knipe, Alden Arthur. Everybody's Washington. Dodd, 1931. . .60

1 86. Komroff, Manuel. Coronet. Grosset. 1930 97

187. Krapp, George Philip. America; the Great Adventure. Knopf,

19*4 75

1 88. Kurlbaum, Margarete Siebert. Mary y Ofyteen of Scots (translat-
ed by Mary A. Hamilton). Harcourt, 1929 53

189. Lacoste, Jean Rene. Lacoste on Tennis. Morrow, 1928 80

190. Lamb, Harold. The Crusades. Doubleday, 1930 54

191. Lang, Andrew. The Conquest of Montezuma's Empire. Long-
mans, 1928 44

192. Lavarre, William J. Up the Mazaruni for Diamonds. Jones,

1922 81

193. Lawton, Mary. Schumann-Heink y the Last of the Titans. Mac-
millan, 1928 1.15

194. Lee, Ettie. Silas Marner (adaptation). Macmillan, 1928 1 .24

195. Leonard, Jonathan Norton. Loki; the Life of Charles Proteus
Steinmetz. Doubleday, 1929 76

196. Lewis, Sinclair. Babbitt. Harcourt, 1922 66

197. Leys, James Farquarson, Jr After You, Magellan! Century,

1927 28

198. Lighty, Kent and Margaret. Shanty-Boat. Century, 1930 60

199. Lindbergh, Charles A. We* Putnam, 1928 64

200. Lippman, Walter. A Preface to Morals. Macmillan, 1929 56

aoi. Lipton, Sir Thomas. Liptori s Autobiography . Duffield, 1932.. .55

202. Lisitzky, Gene. Thomas Jefferson. Viking, 1933 66

203. London, Jack. The Call of the Wild. Grosset, 1903 74

204. London, Jack. Cruise of the Snark. Donohue, 1908 98

205. Lovelace, Delos W. Rockne of Notre Dame. Putnam, 1931 67

206. Lovelace, Maude Hart. Petticoat Court. Day, 1930 78

346 WHAT MAKES A BOOK READABLE

TABLE LXXXV Continued

Predicted

No. Book Index of

Difficulty

207. Lowman, Guy Sumner. Practical Football^ and How To Teach

It. Barnes, 1927 95

208. Lucas, William Palmer. The Health of the Runabout Child; the
Journey from His Mother's Lap to the School Gate. Macmillan,

1923 60

209. Ludwig, EmiL July '14. Putnam, 1929 42

210. MacGregor, Mary. The Story of France. Stokes, 1920 70

211. Madsen, Alfred S., and Lukowitz, J. J. Problems in Furniture
Design and Construction. Bruce, 1928 93

212. Magoffin, R. V. D., and Davis, Emily. Magic Spades, the Ro-
mance of Archaeology. Holt, 1929 58

213. Major, Charles. When Knighthood Was In Flower. Grosset,

1898 70

214. Manning, Sybilla, and Donaldson, A. M. Fundamentals of
Dress Construction. Macmillan, 1926 1.19

215. Markey, Morris. This Country of Yours. Little, 1932 81

216. Marshall, Henrietta Elizabeth. An Island Story. Stokes, 1920 1.07

217. Martin, Harry Brownlow. Whafs Wrong with Your Game?
Dodd, 1930 88

218. Martin, Martha Evans. The Friendly Stars. Harper, 1907 71

219. Martini, Herbert E. Color. Bridgman, 1930 99

220. Masefield, John. Gallipoli. Macmillan, 1917 72

221. Mason, Caroline Atwater. The Spell of Southern Shores; or y
From Sea to Sea in Italy. Page, 1915 47

222. Mason, Gregory. Columbus Came Late. Century, 1931 41

223. Maternity Center Association of New York City. Maternity
Handbook. Putnam, 1932 1.16

224. Maurois, Andr6. Disraeli. Lane, 1927 31

225. Maynard, Theodore. De Soto and the Conquistadores. Long-
mans, 1930 62

226. McBride, Robert Medill. Romantic Czechoslovakia. McBride,

1930 03

227. McKready, Kelvin. A Beginner's Star-Book. Putnam, 1923. . .42

228. McMahon, John R. Wright Brothers. Little, 1930 48

229. Meadowcroft, William H. The Boy's Life of Edison. Harper,

1921 58

230. Melville, Herman. Moby Dick. Modern Library, 1926 28

231. Miller, Janet. Jungles Preferred. Houghton, 1931 71

232. Mills, Dorothy. The Book of the Ancient Greeks. Putnam, 1925 .68

233. Minnigerode, Meade. The Fabulous Forties. Putnam, 1924. . . .41

234. Mitchell, Lucy Sprague. Horses^ Now and Long Ago. Harcourt,

1926 i .08

235. Mitchell, S, Weir. Hugh Wynne. Burt, 1897 78

APPENDIXES 347

TABLE LXXXV Continued

Predicted

No, Book Index of

Difficulty

236. Montgomery, L. M. Anne of Green Gables. Page, 1908 84

237. Moore, Charles W. The Mental Side of Golf. Liveright, 1929. . ,70

238. Morton, Henry C. V. The Call of England. McBride, 1933. . . .78

239. Morton, Henry C. V* In Search of Scotland. Dodd, 1930 51

240. Mowrer, Edgar Ansel. Germany Puts the Clock Back. Morrow,

1933 14

241. Muller, Charles G. How They Carried the Goods. Harcourt,

I93 2 62

242. Munroe, Kirk. The Flamingo Feather. Harper, 1923 48

243. Nisenson, Samuel, and Parker, Alfred. Minute Biographies.
Grosser,, 1931 64

244. NordhoiF, Charles, and Hall, James N. Men against the Sea.
Little, 1934 48

245. O'Brien, John S. By Dog Sled for Byrd; 1,600 Miles across Ant-
arctic Ice. Rockwell, 1931 75

246. Okey, Thomas. The Art of Basket-Making. Pitman, 1912 68

247. Overs treet, Harry Allen. About Ourselves. Norton, 1927 91

248. Parker, Cornelia Stratton. German Summer. Liveright, 1932. . .55

249. Parsons, Geoffrey. The Land of Fair Play. Scribner, 1919 1 .02

250. Parsons, Geoffrey. The Stream of History. Scribner, 1928 68

251. Partridge, Bellamy. Amundsen^ the Bf Undid Norseman.
Stokes, 1929 77

252. Patri, Angelo. School and Home. Appleton, 1925 i . 1 1

253. Perkins, Lucy F. Aesop's Fables. Stokes, 1908 76

254. Phelan, Vincent. The Care and Repair of the Home. Doubleday,

1931 9 1

255. Picken, Mary Brooks. How To Make Draperies (Singer Sewing
Library, No. 4). Singer Sewing Machine Co., 1930 91

256. Poe, Edgar Allen. The Gold Bug (adapted as The Gold Insect in
basic English). Trubner, 1932 81

257. Polk, Ralph W. The Practice of Printing. Manual Arts, 1926. . .96

258. Powell, E. Alexander. Undiscovered Europe. Washburn, 1932. .15

259. Pupin, Michael. From Immigrant to Inventor. Scribner, 1925. . .60

260. Rawlings, Marjorie K. South Moon Under. Scribner, 1933. . . .85

261. Remarque, Erich Maria. All Quiet on the Western Front. Gros-

set, 1928 80

262. Repplier, Agnes. P2rt Marquette. Doubleday, 1929 68

263. Richardson, William L., and Owen, Jesse M. Literature of the
World. Ginn, 1922 64

264. Rinehart, Mary Roberts. Nomad's Land. Doran, 1922 66

265. Rinehart, Mary Roberts, Tenting To-Night, Doubleday, 1928 .89

266. Robert, Henry M. Roterfs Rules of Order Revised. Scott, Fores-
man, 1915 75

348 WHAT MAKES A BOOK READABLE

TABLE LXXXV Continued

Predicted

No. Book Index of

Difficulty

267. Robinson, James Harvey. The Mind in the Making. Harper,

I9 21 43

268. Robinson, Will H. Under Turquoise Skies; Americas South-
west from the Days of the Ancient Cliff-Dwellers to Modern Times.
Macmillan, 1928 34

269. Roche, Mazo de la. Jalna. Little, 1928 68

270. Rockwell, Frederick Frye. Gardening under Glass. De la Mare,

I9 28 44

271. Rogers, Agnes, and Allen, F. L. The American Procession. Har-

per, 1933 15

272. Rolland, Romain. Jean-Christophe. Holt, 1910 23

273. Rolvaag, O. E. Giants in the Earth. Harper, 1929 69

274. Roosevelt, Franklin D. Looking Forward. Day, 1933 70

275. Roosevelt, Theodore. Stories of the Great West. Century, 1909 .60

276. Ross, Leland M., and Grobin, A. W. This Democratic Roosevelt;

the Life Story of "F.D" Dutton, 1932 53

277. Rush, Mary Wheeler. The Ignoramus Garden Book; a Practical
Handbook for the Beginner. Sears, 1931 73

278. Russell, Phillips. John Paul Jones. Brentano's, 1927 42

279. Sabatini, Rafael. The Banner of the Bull. Grosset, 1927 61

280. Sabatini, Rafael. Scaramouche. Grosset, 1921 82

281. Sandburg, Carl. Abe Lincoln Grows Up. Harcourt, 1926 58

282. Sandburg, Carl, and Angle, Paul M. Mary Lincoln. Harcourt,

1932 _ s 2

283. Sawyer, Robert V., and Perkins, Edwin. Water Gardens and
Goldfish. De la Mare, 1928 60

284. Seabrook, William B. Jungle Ways. Harcourt, 1931 59

285. Sedgwick, Henry D wight. France; a Short History of Its Poli-
tics^ Literature, and Art from Earliest Times to the Present.
Little, 1929 33

286. Seed, T. Rutherford. Basket Work; a Practical Handbook. Ox-
ford University Press, 1927 i .00

287. Shay, Frank. Here's Audacity! (American legendary heroes)
Macaulay, 1930 86

288. Sichel, Edith Helen. The Renaissance. (Home University of
Modern Knowledge. No. 27.) Holt, 1914 32

289. Singmaster 3 Elsie. The Book of the Colonies. Doubleday, 1929.. .71

290. Singmaster^ Elsie, The Book of the United States. Doubleday,

1926 69

291. Siringo, Charles A. Riata and Spurs. Houghton, 1927 78

292. Skinner, Otis. Footlights and Spotlights. Bobbs, 1924 55

293. Slocum, Joshua. Sailing Alone around the World. Century,

1900 58

APPENDIXES 349

TABLE LXXXV Continued

Predicted

No. Book Index of

Difficulty

294. Smith, Alfred E. Up to Now; an Autobiography. Viking, 1929 .. .63

295. Smith, Andre". The Scenewright. Macmillan, 1927 53

296. Snedeker, Caroline Dale. The Spartan. Doubleday, 1922 78

297. Soule, George. The Useful Art of Economics. Macmillan, 1929 .60

298. Steele, Fletcher* Design in the Little Garden. Atlantic Monthly
Press, 1924 66

299. Steffens, Lincoln. The Autobiography of Lincoln Steffens. Har-
court, 1931 63

300. Stevenson, Robert Louis. Treasure Island. Scribner, 1921 64

301. Stewart, Cora Wilson. Country Life Readers, Book I. Johnson,

1915 1-84

302. Stewart, Cora Wilson. Country Life Readers, Book II. John-
son, 1916 i .33

303. Stewart, Cora Wilson. Country Life Readers, Book III. John-
son, 1917 1.04

304. Stote, Dorothy. Making the Most of Your Looks. Brentano's,

1926 89

305. Stribling, T. S. Teef tallow. Doubleday, 1926 69

306. Sublette, Clifford M. The Scarlet Cockerel. Little, 1929 43

307. Suckow, Ruth. Bonney Family. Knopf, 1928 86

308. Sullivan, Mark. Our Times; 1900-1925. Scribner, 1930 51

309. Sullivan, Mark. Our Times; the Turn of the Century. Scribner,

19*7 r 4

310. Tappan, Eva March. When Knights Were Bold. Houghton,

1911 66

311. Tarkington, Booth. Penrod. Grosset, 1914 71

312. Thorn, Douglas A. Everyday Problems of the Everyday Child.
Appleton, 1929 36

313. Thomas, Lowell. With Lawrence in Arabia. Garden City, 1924 .39

314. Thomason, John W., Jr. Fix Bayonets! Scribner, 1926 59

315. Tipton, Edna S. Table Decorations for All Occasions. Stokes,

19*4 9

316. Tracy, Louis. The Wings of the Morning. Grosset, 1903 54

317. Trine, Ralph Waldo. In Tune with the Infinite. Bobbs, 1897 , . .86

318. Tschiffely, A. F. Tschiffely's Ride; Ten Thousand Miles in the
Saddle from Southern Cross to Pole Star. Simon, 1933 47

319. Tunney, Gene. A Man Must Fight. Houghton, 1932 75

320. Turgenev, Ivan S. Fathers and Sons. Dutton, 1922 43

321. Twain, Mark. The Adventures of Tom Sawyer. Harper, 1922. . .86

322. Usher, Roland G. The Story of the Great War. Macmillan, 1919 .72

323. Van Loon, Hendrik Willem. America. Boni, 1927 24

324. Van Loon, Hendrik Willem. Ancient Man. Boni, 1920 .67

325. Van Loon, Hendrik Willem, The Story of Mankind. Boni, 1921 .47

350 WHAT MAKES A BOOK READABLE

TABLE LXXXV -Continued

Predicted

No. Book Index of

Difficulty

326. Van Loon, Hendrik Willem. Van Loon's Geography; The Story

of the World We Live /. Simon, 1932 46

327. Van Metre, T. W. Trains^ Tracks and Travel. Harcourt, 1931 i .08

328. Villiers, Alan J. By Way of Cape Horn, Holt, 1930 73

329. Wadhams, Caroline R. Simple Directions for the Chambermaid.
Longmans, 1917 i , u

330. Wain, Nora. The House of Exile. Little, 1933 57

331. Walpole, Hugh. Fortitude. Doubleday, 1913 67

332. Washington, Booker T. Up from Slavery. Doubleday, 1920, . . .72
333- Welzl, Jan. Thirty Years in the Golden North. Macmillan, 1932 .94

334. West, Michael. Robinson Crusoe New Method Readers, Sup-
plementary Reader III. Longmans, 1931 2 . 06

335. Weyman, Stanley J. Under the Red Robe. Grosset, 1923 91

336. Wharton, Edith. Ethan Frome. Scribner, 1911 68

337. White, Stewart Edward. Daniel Boone> Wilderness Scout. Al-

lyn & Bacon, 1926 76

338. White, William C. These Russians. Scribner, 1931 65

339- Wilder, Thornton. The Woman of Andros. Boni, 1930 86

340. Williams, Albert Rhys. The Russian Land. New Republic,

1927 51

341. Willson, Beckles. Canada (Romance of Empire Series). Nelson,

1933 47

342. Winkler, John K. John D.; a Portrait in Oils. Blue Ribbon,

1929 63

343. Winn, Mary Day. The Macadam Trail; Ten Thousand Miles by
Motor Coach. Knopf, 1931 16

344. Wister, Owen. The Seven Ages of Washington. Macmillan, 1917 .08

345. Woodburn, James A., and Moran, Thomas F. The Makers of
America. Longmans, 1922 83

346. Wooley, C. Leonard. Ur of the Chaldees. Scribner, 1930 28

347. Wright, Eugene. The Great Horn Spoon. Bobbs, 1928 53

348. Wright, Harold Bell. The Winning of Barbara Worth. Burt,

1911 83

349. Wyss, David. Swiss Family Robinson. Harper, 1909 34

350. Wyss, David. Swiss Family Robinson; in Words of One Syl-
lable. AltemuSj 1900 70

APPENDIX G

Table LXXXVI lists the reading textbooks for Grades II-IX used in inter-
preting areas of difficulty represented by adult books in terms of elementary
grade levels for which they are structurally appropriate. How the books were
used for this purpose is explained in chapter vi.

TABLE LXXXVI

TEXTBOOKS INT READING USED IN THE INTERPRETATION OF AREAS OF
DIFFICULTY REPRESENTED BY 350 ADULT BOOKS

Bolenius, Emma Miller. Literature in the Junior High School, Books I, II, and

III. Houghton, 1926-28.
Bryce, Catharine T.; Hardy, Rose Lees; and Turpin, Edna. Newson Readers,

Books II, III, IV, V, and VI. Newson, 1927-29.
Coleman, Bessie; Uhl, Willis; and Hosic, James. The Pathway to Reading,

Books II, III, IV, V, and VI. Silver, Burdett, 1925-26.
ELSON, William H.; Gray, William S.; and Keck, Christine M. Ehon Basic

Readers, Books II, III, IV, V, VI. Scott, Foresman, 1931.
Engleman, J. O., and McTurnan, Lawrence. Guide Books to Literature, Books

I, II, and III. Laidlaw, 1925-26.

Freeman, Frank N.; Storm, Grace E.; Johnson, Eleanor M.; and French,

W. C. Child Story Readers, Books II, III, IV, V, and VI. Lyons & Carna-

han, 192729.
Gates, Arthur L; Huber, Miriam Blanton; and Ayer, Jean Y. The Work-Play

Books, Books II, III, IV, V, and VI. Macmillan, 1930-32.
Greenlaw, Edwin; Miles, Dudley; Stratton, Clarence; and Keck, Christine M.

Literature and Life, Books I, II, and III. Scott, Foresman, 1929-33.
Haggerty, Melvin E., and Smith, Dora V. Reading and Literature^ Books I,

II, and III. World Book Co., 1927-28.

Hardy, Marjorie. New Stories, Book II; Best Stories, Book III. Wheeler,

1926-27.
Hill, Howard C.; Lyman, Rollo L.; and Moore, Nelle E. Reading and Living,

Books I, II, and III. Scribner, 1930.
Ringer; Edith Hope; Sewell, J. W.; Harris, Albert Mason; Stockton, Helen

M.; and Downie, Lou Chase. Citizenship Readers, Books II, III, IV, V,

VI. Lippincott, 1930.
Lewis, Wm. D.; Rowland, A. L.; Marshall and Gehres. New Silent Readers,

Books II, III, IV, V, VI, VII, and VIIL Winston, 1930-31.
Lyman, Rollo, and Hill, Howard C. Literature and Living, Books I and II.

Scribner, 1925.
Manly, J. M.; Rickert, Edith; and Leubric, Nina. Good Reading, Books II,

III, IV, V, VI. Scribner, 1926-28.

352 WHAT MAKES A BOOK READABLE

TABLE LXXXVICe>wft*/

Patterson, Samuel White. Bobls-Merrill Literature Series, Books I, II, and

III. Bobbs, 1928.

Pennell, Mary, and Cusack, Alice. The Children's Own Readers, Books II, III,

IV, V, and VI. Ginn, 1929.

Ross, J. M., and Schweikert, H. C. Adventures in Literature^ Books VII,

VIII, and IX. Harcourt, 1927-28.
Suzzallo, H.; Freeland, G.; McLaughlin, K. L.; and Skinner, A. Fact and

Story Readers, Books II, III, IV, V, VI, VII, and VIII. American Book,

Theisen, Wm. W., and Leonard, S. A. Real Life Stories, Books VII and VIII.
Macmillan, 1929.

INDEX

Adjacent areas of difficulty, significance
of, 208

Adult books studied, 195

Adult readers, percentage of, reading
books in different areas of difficulty,
220

Adult reading, grade levels of, 76, 78

Adult reading ability: based on informa-
tion concerning last year in school,
239; determined by expressed prefer-
ence for materials, 243 ; determined by
knowledge of materials previously
read, 238; use of objective tests in de-
termining, 236

Adult reading materials: differences in,
with respect to elements of difficulty,
143; difficulty of, 194; diversity of
vocabulary in, 153; elements influ-
encing difficulty of, 94; length of sen-
tences in, 155; monosyllables in, 151;
samples of, arranged in order of diffi-
culty, 244, 245, 247, 248; simplicity
of sentence structure in, 1 56

Adult reading test, description of, 59, 60

American Association for Adult Educa-
tion, v, vii

American Library Association, v, vii

An average reading score used in pre-
dicting difficulties, 196

Areas of difficulty: books characterizing,
209; meaning of, in terms of grade-
levels of reading ability, 213

Areas of difficulty defined, 204

Average length of sentences in books, 1 88;
in general magazines, 168; in news-
papers, 1 80

Average reading scores, predicted for
eighty-one textbooks in reading, ai6

Asides, appositives, and parenthetical ex-
pressions as indicators of difficulty, 106

Ayer, Adelaide M., 107

Ayer, N. W., and Sons, 144

Better reading material, need for simple

presentation of, 264
Boas, Ralph Philip, in

Books: classified according to simplicity
of sentence form, r88; classified ac-
cording to vocabulary diversity, 186;
differences in length of word, 184;
relative difficulty according to aver-
age length of sentences, 188

Books analyzed, 147, 148, 149

Books, difficulty of composite ranking,
191

Burch, Mary Crowell, -234

Carnegie Corporation of New York, vii

Chancellor, John, vii, 24, 34, 38, 233, 249

Cheney, O. H., 229, 262

Classifying reading material, by predicted
index of difficulty, 253

Cleary, Margaret D. 3 72

Committee on the Reading Interests and
Habits of Adults, v

Comprehension of general reading ma-
terial, how to measure, 58

Conrad, L. R., 24

Content, general aspects of, the relative
importance of, 35

Content as influencing readability, 33

Content and structural words as indica-
tors of difficulty, 104

Cornell, Ethel, 276

Cornell Reading Vocabulary for Foreign-
born Adults, 276

Criterion of difficulty, obtaining, 96

Dale, Edgar, vii, 100, 101, 102, 236, 275

DeVoss, J. C., 257

Dickerson, L. L., 249

Different words, as indicators of difficulty,
102

Difficulty: areas of defined, 204* ele-
ments in test paragraphs influenc-
ing, 96; predicted in terms of an aver-
age reading score, 196; range of, for
350 books, 200

Difficulty of adult reading materials, ele-
ments influencing, 94

Difficulty of newspapers, with respect to
elements of difficulty, 178

355

356

WHAT MAKES A BOOK READABLE

Difficulty of reading materials,, as related

to structural elements, 7
Direct and indirect discourse as indicators

of difficulty, 1 1 1
Diversity of vocabulary in adult reading

materials, 153
Dolch, Edward William, 102, 105

Easy and hard words, as indicators of
difficulty, 100

Elements in general reading of use in esti-
mating difficulty, 126

Elements in test paragraphs that influence
difficulty, 96

Elements of difficulty: correlation with
criterion of difficulty, 114; in differ-
ent classes of newspapers, 179; selec-
tion of for experimentation, 113; selec-
tion of most useful for prediction, 130;
significance of for all readers, 121; sig-
nificance of for "best" readers, 122;
significance of for "poorest" readers,

123
Elements of difficulty in general reading

material, for different classes of read-
ers, 117
Elements, use of in estimating difficulty,

131
Expressional elements, list of, influencing

difficulty, 97

Factors influencing readability: a pos-
sible list of, 25; four major categories
of, 26; judgment concerning influ-
ence of, 30; nature of replies concern-
ing, 29; technique of obtaining opin-
ion relative to, 28

Faucett, Lawrence, 277

Faucett's Minimum Vocabulary for Read-
ing and Understanding, 277

Federal Emergency Relief Commission,
vii

Felsenthal, Emma, 23

Figures of speech as indicators of diffi-
culty, no

Findings, treatment of, 151

Findings of report, applicability of, 9

Findings of study, compared with those
of other studies, 124

Flexner, Jennie M., vii

Format; factors of in a readable book,

40; general aspects of, the relative
importance of, 42
Franz, S. L, 41

Gates, Arthur I., 276

Gates's Reading Vocabulary for Primary
Grades, 276

General magazines, classified by: aver-
age length of sentences, 168; percent-
age of different words, 166; percentage
of monosyllables, 159; percentage of
simple sentences, 171

General magazines, difficulty of, com-
posite ranking, 175, 176

General reading material: how to measure
comprehension of, 58; how well adults
read, 67

Grade levels of adult reading, 76

Gray, William S., 95, 240, 249, 274

Gray, Wil Lou, 240, 249

Griffing, H., 4 i

Groups tested, 61, 74

Harris, Helen C., 107

Hoit, Doris, vii, 196

Holzinger, Karl J., 83, 126, 136, 137, 163,

205, 206

How to prepare readable materials, 261
How to select readable materials, 224
How well adults read, 57

Image-bearing and non-image-bearing
words, as indicators of difficulty, 104

Inter-correlations of elements of difficulty,
128, 129

International Kindergarten Union: Child
Study Committee of, 101 ; Word List
of, 101

Investigation of readability, steps of pro-
cedure in, 10

Johnson, George R., 103, 255

Kelley, Truman L., 137
Kelly, F.J., 257

L,,H.W.,i5o

LaBrant, Lou L., 1 50

Larsen, Laura M., 235

Length and kind of sentences, as indi-
cators of difficulty, 107

Length of selection, as indicator of diffi-
culty, no

INDEX

357

Length of sentences, in adult reading ma-
terials, 155

Lewerenz, Alfred S., 105, 254, 270

Library readers, methods of obtaining
reactions from, 45

Limitations of the study, 94, 95

Lively, Bertha A,, 102

Lubbock, Percy, in

Magazines analyzed, 144, 145, 146
Major findings of study, applications of,

Maki, Itsu, 277
Malinowski, Bronislaw, no
Mason, Charles W., 231
McAdams, Mary Ann, 234
McCluskey, Howard Y., 103
Miers, Sir Henry A., 237
Monosyllables in adult reading materials,

151

Monroe, Ruth, 95, 274
Monroe, W. S., 257

Negro adults tested, reading achievement
of, 91

Newspapers: differences in difficulty in
departments of, 183; occurrence of
significant elements of difficulty in,
179; variations with respect to ele-
ments of difficulty, 178

Newspapers analyzed, 146, 147, 148

Ogden, C. K., 38, 106, no, 278, 279

Ogden's Basic English Vocabulary, 278

Ojemann, R. H., 102, 240

Organization, factors of, influencing read-
ability, 43

Organization, general aspects of, relative
importance of, 45

Orndorff, Bernice, 108

Parts of speech, as indicators of difficulty,
103

Personal pronouns, as indicators of diffi-
culty, 103

Peterson, Mildred O., 228

Phrases, as indicators of difficulty, 109

Potential indicators of difficulty, in adult
reading material, 98

Predication, as an indicator of difficulty,
109

Predicted scores, verification of, 139

Predicting difficulty, in terms of an aver-
age reading score, 134

Prediction of difficulty, by a smaller num-
ber of elements, 137; reliability of, 136

Preparing readable books, points to be
considered in, 269

Preparing readable materials, contribu-
tions of scientific studies to, 272

Pressey, L. C., 255

Pressey, S. L., 102

Problems for further study, suggestions
of, 19

Properties of paragraphs, that are po-
tential indicators of difficulty, 99

Properties of sentences, that are poten-
tial indicators of difficulty, 99

Properties of words, that are potential
indicators of difficulty, 98

Ranking reading materials for relative
difficulty, 254

Readable book: approach to definition of,
5; characteristics as described in cur-
rent literature, 22; dictionary defini-
tion of, 22; factors of format in, 40

Readable books: many kinds of, 261;
number available for adults of limited
ability, 262; selection of, approach to
study of, 233

Readable materials: how to prepare, 261;
interest in at present, I

Readable writing, problems of, 283

Readable writing for adults, tentative
standards for, 288

Readability: factors of content that in-
fluence, 33; factors of organization in-
fluencing, 43; factors of style con-
tributing to, 36; possible factors in-
fluencing, 25; relation of vocabulary
to, 275 ; summary of inquiry relative

to, 54

Readers, kinds reporting, 47
Readers* reactions, an analysis of, 48
Reading ability, nature of measure of, 6
Reading ability of adults, summary of

testing program, 92

Reading achievement: comparison on two
tests of, 82; of negro adults, 91; of
village groups, 86; relation to amount
of educational training, 84
Reading Habits Committee, of the Amer-
ican Association of Adult Education

358

WHAT MAKES A BOOK READABLE

and of the American Library Associa-
tion, vii

Reading interests and habits of adults,
what has been discovered about, 274

Reading materials for adults, how to se-
lect, 224

Reading materials, selection of, aspects
to be considered in, 230

Reading passages, differences in, with one
element held constant, 281, 282

Rejall, Alfred E., 277

Rejall Word List for Voters in Citizen-
ship, 277

Relative difficulty: defining with respect
to specific elements in general maga-
zines, 157; procedure in estimating,
150

Report, what it is about, i

Richards, I. A., 106, no

Robinson, James Harvey, 22

Sampling of books, adequacy of, 140

Scaling materials for difficulty, 256

Scudder, Harold H., 148

Selecting reading materials of appropriate
difficulty, 250

Selecting right books for reader, a typical
situation, 234

Selection of adult reading materials, im-
portance of, 224

Simplicity of sentence form in books, 188;
in general magazines, 171; in news-
papers, 1 8 1

Simplicity of structure, in adult reading
materials, 156

Snedden, David, 227

Stote, Dorothy, 199

Structural elements, reasons for investi-
gation of, 7

Style, general aspects of, the relative im-
portance of, 38

Stylistic devices, as indicators of diffi-
culty, 112.

Sub-committee on readable books, v, 23

Subordinate clauses, as indicators of diffi-
culty, 109

Swenson, Elaine, 276^278

Swenson's Minimum Vocabulary for
Foreigners, 276

Swineford, Frances, vii

Syllabic length of word, as indicator of
difficulty, 103

Syllabic word-length, differences in maga-
zines of different content, 163; differ-
ences among departments of news-
papers, 183

Technique of paragraph development, as
indicator of difficulty, 112

Testing adult reading ability, securing
groups for, 60

Thompson, Ruth Culver, 108

Thorndike, Edward L., 101, 108, 275

Thorndike Word List, 275

Thorndike's Teachers Word Book, 101

Til ton, J. Warren, 240, 249

Tyler, Ralph W., vii, 35, ico, 236, 274

Types of narration, as indicators of diffi-
culty, in

Tyson, Levering, 228

Vernon, Madeline D., 41

Village group tested, achievement of, 86

Vocabulary diversity of books, 186; of

general magazines, 166; of newspapers,

179
Vogel, Mabel, 102

Waples, Douglas, vii, 24, 35, 274, 275

Washburne, Carleton, 102, 236

West, Michael, vii, 244, 266, 278, 281

Wiggins, G. T., 76

Wilson, Louis R., vii

Winslow, Amy, 231

Witty, Paul A., 150

Words associated with adult living, as
indicators of difficulty, 105

Words beginning with w, h y b, i, or e, as
indicators of difficulty, 104

Words expressing abstractions, as indi-
cators of difficulty, 105

Words not known to 90 per cent of sixth-
grade pupils, 101

PRINTED
I IN U-S-Aj

The University of Chicago

STUDIES IN LIBRARY SCIENCE

Under the General Editorship of
THE GRADUATE LIBRARY SCHOOL

AN INTRODUCTION TO LIBRARY
SCIENCE

By PIERCE BUTLER

A stimulating discussion of "The Sociological Problem," "The Psycho-
logical Problem," and "The Historical Problem" in librarianship as a social
science. Also a chapter on "Practical Considerations." 1 18 pages, cloth, $1.50

WHAT MAKES A BOOK READABLE

By WILLIAM S. GRAY and BERNICE E. LEARY

An analysis of the elements of difficulty and of interest in various reading
materials, with suggestions for proper selection of books for adults. A valu-
able study for educators, librarians, publishers, writers, and workers in adult
education. 390 pages, cloth, $3.00

THE GOVERNMENT OF THE
AMERICAN PUBLIC LIBRARY

By CARLETON B* JOECKEL

A study of the position of the public library in the structure of government
in the United States. Deals with legal forms and types of governmental
organization of the public library and its relation to the city or other political
unit to which it is attached or which it serves. Considers the library pri-
marily as a piece of governmental machinery and its efficiency as such,
rather than the practical problem of its internal administration. A book for
students of government as well as library administrators.

420 pages, cloth, $3.00

The University of Chicago Press

33302